Home

OMNeT++ - Fossies

1. case FSM_Enter SLEEP schedule end of s scheduleAt simTime break leep period startStopBurst case FSM_Exit SLEEP schedule end of t scheduleAt simTime 4 tr if m e FSM_G break his burst exponential burstTimeMean startStopBurst start of next burst texponential sleepTimeMean ansition to ACTIVE state sg startStopBurst rror invalid event in state ACTIVE oto fsm ACTIVE case FSM_Enter ACTIVE sc sched break case FS transition to either S if m hedule next sending uleAt simTin tA M_Exit ACTIVE END or SLEE ne texponential sendIATime sendMessage G sg sendMessage Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 84 FSM_Goto fsm SEND else if msg startStopBurst cancelEvent sendMessage FSM_Goto fsm SLEEP else error invalid event in state ACTIVE break case FSM_Exit SEND generate and send out job char msgname 32 sprintf msgname job d i ev lt lt Generating lt lt msgname lt lt endl cMessage job new cMessage msgname job gt setLength long msgLength job gt setTimestamp send job out
2. return to ACTIV FSM_Goto fsm ACTIV break CM CI 5 5 Message transmission modeling Data rate modeling If data rate is specified for a connection a message will have a certain nonzero transmission time depending on its length This means that when a message is sent out through an output gate the message reserves the gate for a given period it is being transmitted 28 800 bps Fig 5 1 Connection with a data rate While a message is under transmission other messages have to wait until the transmission finishes You can still use send while the gate is busy but the message s arrival will be delayed just like the gate had an internal queue for the messages waiting to be transmitted The OMNeT class library provides you with functions to check whether a certain output gate is transmitting or to learn when it finishes transmission If the connection with a data rate is not the immediate one connected to the simple module s output gate but the second one in the route you have to check the second gate s busy condition Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 85 Implementation of message sending Message sending is implemented in the following way the arrival time and the bit error flag of a message are calculated at once when the send or similar function is invoked That is if the message travels through several links until
3. foo simulation systemModule gt par foo op_prg_odb_print_minor ev lt lt hello lt lt endl op_prg_odb_print_major ev printf Ssimulation error Your fault error d ec op_subq_ Create a queue object and then manipulate it with its member functions cQueue queue queue insert msg if queue empty msg queue pop List op_prg_list_ cLinkedList list list insert ptr if list empty ptr list pop Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix A OPNET and OMNeT 193 Topology The cTopology class offers similar functionality and you op_rte_ can expect greater speed than with OPNET s routing functions Packet op_pk_create op_pk_destroy packet fields op_pk_nfd_set op_pk_nfd_get_ op_pk_fd_set op_pk_fd_get Use the cMessage class cMessage msg new cMessage delete msg Message parameters A parameter has both name and index msg gt par foo foo msg gt addPar new foo foo int foo msg gt par foo int fooindex msg gt parList find foo msg gt par fooindex foo packet field modelled size msg gt addLength Message parameters do not have associated modelled bit sizes Message length can be used instead msg gt addPar dest_addr dest_addr 32
4. Player 1 SmartPlayer to move SmartPlayer is taking 1 out of 27 sticks Player 1 SmartPlayer took 1 stick s Sticks left 26 eea Sticks left 5 Technical University of Budapest Dept of Telecommunications OMNeT Manual An Example The NIM Game 33 Player 1 SmartPlayer to move SmartPlayer is taking 4 out of 5 sticks Player 1 SmartPlayer took 4 stick s Sticks left 1 Player 2 SimplePlayer to move SimplePlayer is taking 1 out of 1 sticks Player 2 SimplePlayer took 1 stick s Player 1 SmartPlayer won lt gt Module nim game Simulation stopped with endSimulation End run of OMNeT 3 4 Other examples An enhanced version of the NIM example can be found among the sample programs It adds a third interactive player and derives specific player types from a Player abstract class It also adds the possibility that actual types for player1l and player2 can be specified in the ini file or interactively entered by the user at the beginning of the simulation Nim does not show very much of how complex algorithms like communication protocols can be implemented in OMNeT To have an idea about that look at the Token Ring example It is also extensively commented though you may need to peep into the user manual to fully understand it Other programs in the example manual are Dyna and FDDI Dyna models a simple client server network and demons
5. 15 12 7 Class cEqdHistogramBase Base class for equal cell size histograms 15 12 8 class cLongHistogram cLongHistogram is derived from cEgqdHistogramBase which contains most of the functionality The histogram is set up in the following way e the cell size is always integer 1 2 3 etc Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 251 e rangemin rangemax and the cell boundaries are at halves For example 1 5 10 5 This is done so to prevent misunderstandings e the number of cells is exactly num_cells e the range_ext_factor is also kept The actual histogram range will be min max extended range_ext_factor times and rounded up to the nearest integer multiple of num_cells Construction destruction copying cLongHistogram cLongHistogram amp r Copy constructor cLongHistogram const char s NULL unsigned num_cells 10 Constructor that takes the object name and the number of cells virtual cLongHistogram Destructor virtual cObject dup cLongHistogram amp operator cLongHistogram res Duplication and assignment work all right with cLlongHistogram Redefined virtual functions virtual const char className j Returns a pointer to the class name string CcLongHistogram virtual void info char buf virtual const char inspectorFactoryName Redefined virtual void transform If the result co
6. Define Network macros occur in the code generated by the NEDC compiler List of available module types A cModuletType object knows how to create a module of a specific type If is compound it holds a pointer to a function that can build up the inside Usually Define_Module macros for compound modules occur in the code generated by the NEDC compiler for simple modules the Define_Module lines are added by the user cHead classes Register_Class ClassRegister List of available classes of which the user can create an instance A cClassRegister object knows how to create an empty object of a specific class The list is used by the creat eOne function that can create an object of any registered type from a string containing the class name E g ptr createOne cArray creates an empty array createOne is used by the PVM extension Register_Class macros are present in the simulation source files for existing classes has to be written by the user for new classes cHead functions cHead linktypes Define_Function cFunctionType Define_Link cLinkType List of mathematical functions A cFunctionType object holds a pointer to the function and knows how many arguments it takes List of link types A cLinkType object knows how to create cPar objects representing the delay error and datarate attributes for a channel Define_Link macr
7. e cTransientDetection base class for transient detection e cAccuracyDetection base class for result accuracy detection Basic usage TBD comments Attaching detection objects to a cStatistic and getting pointers to the attached objects addTransientDetection cTransientDetection object addAccuracyDetection cAccuracyDetection object cTransientDetection transientDetectionObject cAccuracyDetection accuracyDetectionObject Detecting the end of the period e polling the detect function of the object e installing a post detect function Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 134 Transient detection Currently one transient detection algorithm is implemented i e there s one class derived from cTransientDetection The cTIDExpandingWindows class uses the sliding window approach with two windows and checks the difference of the two averages to see if the transient period is over void setParameters int reps 3 int minw 4 double wind 1 3 double acc 0 3 Accuracy detection Currently one transient detection algorithm is implemented i e there s one class derived from cAccuracyDetection The algorithm implemented in the cADByStddev class is divide the standard deviation by the square of the number of values and check if this is small enough void setParameters double acc 0 1 int reps 3 6 14 Recordi
8. setRange lower upper setRangeAuto num_firstvals range_ext_factor setRangeAutoLower upper num_firstvals range_ext_factor setRangeAutoUpper lower num_firstvals range_ext_factor setNumFirstVals num_firstvals The following example creates a histogram with 20 cells and automatic range estimation cDoubleHistogram histogram histogram 20 histogram setRangeAuto 100 1 5 Here 20 is the number of cells not including the underflow overflow cells see later and 100 is the number of observations to be collected before setting up the cells 1 5 is the range extension factor It means that the actual range of the initial observations will be expanded 1 5 times and this expanded range will be used to lay out the cells This method increases the chance that further observations fall in one of the cells and not outside the histogram range range of initial observations histogram range Fig 5 4 Setting up a histogram s range Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 129 After the cells have been set up collecting can go on The transformed function returns true when the cells have already been set up You can force range estimation and setting up the cells by calling the transform function The observations that fall outside the histogram range will be counted as underflows and overflows The number of underflows an
9. e other variables which belong to the state of the module retry counts packet queues etc e values retrieved computed once and then stored values of module parameters gate indices routing information etc e pointers of message objects created once and then reused for timers timeouts etc e variables objects for statistics collection You can initialize these variables from the initialize function The constructor is not a very good place for this purpose because it is called in the network setup phase when the model is still under construction so a lot of information you may want to use is not yet available then Another task you have to do in initialize is to schedule initial event s which trigger the first call s to handleMessage After the first call handleMessage must take care to schedule further events for itself so that the chain is not broken Scheduling events is not necessary if your module only has to react to messages coming from other modules finish is used in the normal way to record statistics information accumulated in data members of the class at the end of the simulation Application area There are two areas where handleMessage is definitely a better choice than activity 1 For modules which have to maintain little or no state information such as packet sinks 2 Other good candidates are modules with a large state space and many arbitrary state transition possibilities i e where the
10. Coroutine stack size All the simulation programmer needs to care about coroutines is to choose the processor stack size for them This cannot be automated Eerrr at least not without hardware support some trick with virtual memory handling 8 or 16 kbytes is usually a good choice but you may need more if the module uses recursive functions or has local variables which occupy a lot of stack space OMNeT has a built mechanism that will usually detect if the module stack is too small and overflows OMNeT can also tell you how much stack space a module actually uses so you can find it out if you overestimated the stack needs initialize and finish with activity0 Because local variables of activity are preserved across events you can store everything state information packet buffers etc in them Local variables can be initialized at the top of the activity function so there isn t much need to useinitialize However you need finish if you want to write statistics at the end of the simulation And because finish cannot access the local variables of activity you have to put the variables and objects that contain the statistics into the module class You still dont need initialize because class members can also be initialized at the top of activity Thus a typical setup looks like this pseudocode Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 73
11. Opens the statistical output file with the currently set fileName void closeFile Closes the statistical output file long getNewID Returns a unique ID for constructing a new cOutVector object The user should not explicitly call this function FILE getHandle Returns the file pointer of the statistical output file 15 12 Statistical data collection cStatistic and derived classes Header file cstat h sim directory The statistical data collection classes cStatistic cStdDev cWeightedStdDev cDensityEstBase cHistogramBase cVarHistogram cEQDHistogramBase cLongHistogram cDoubleHistogram cPSquare and cKSplit Several of those are abstract classes The ones which are not are cStdDev cVarHistogram cLongHistogram cDoubleHistogram cPSquare and cKSplit Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 246 cObject cHistogramBase Fig 4 1 Hierarchy of statistics collection classes cWeightedStdDev cKSplit 15 12 1 class cStatistic cStatistic is the base class for all statistical data collection classes cStatistic itself adds no data members or algorithms to cObject it only defines virtual functions that will be redefined in descendants No instance of cStat istic can be created Construction destruction copying cStatistic cStatistic amp r Copy co
12. burst_ctr burst_length schedule first packet of first burst scheduleAt simTime new cMessage void BurstyGenerator handleMessage cMessage msg generate amp send packet cMessage pkt new cMessage send pkt out Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 78 if this was the last packet of the burst if burst_ctr 0 schedule next burst burst_ctr burst_length scheduleAt simTime exponential 5 0 msg else schedule next sending within burst scheduleAt simTime exponential 1 0 msg Advantages and drawbacks Advantages e consumes less memory no separate stack needed for simple modules e fast function call is faster than switching between coroutines Drawbacks e local variables cannot be used to store state information e need to redefine initialize e programming model is inconvenient in some cases Other simulators Many simulation packages use a similar approach often topped with something like a state machine FSM which hides the underlying function calls Such systems are e OPNET MIL3 Inc which uses FSM s designed using a graphical editor e NetSim clones OPNET s approach e SMURPH University of Alberta defines a somewhat eclectic language to describe FSMs and uses a precompiler to turn it into C code e Ptolemy UC Berkeley uses a similar method OMNeT s
13. Modules have input and output gates the tiny boxes labeled in out from_playerl etc in the figure An input and an output gate can be connected connections or links are shown as in the figure as arrows During the simulation modules communicate by sending messages through the connections Technical University of Budapest Dept of Telecommunications OMNeT Manual An Example The NIM Game 28 The user defines the topology of the network in NED files We placed the model description in two files the first file defines the simple module types and the second one the system module The first file NED keywords are typed in boldface file nim_mod ned Simple modules in nim ned Declaration of simple module type Game simple Game parameters num_sticks initial number of sticks first_move 1 Playerl 2 Player2 gates in from_playerl input and output gates from_player2 for connecting to Playerl Player2 out to_playerl to_player2 endsimple Now the declarations of the two simple module types Any one of the two types can be Playerl or Player2 A player that makes random moves simple SimplePlayer gates in in gates for connecting to Game out out endsimple A player who knows the winning algorithm simple SmartPlayer gates in in gates for connecting to Game out out endsimple The second file fil
14. To be able to use makemake you have to collect all your sources ned cc h files in one directory Large models which spread across several directories are covered later in this section Then type makemake If you already had a makefile in that directory you ll have to force makemake to overwrite it makemake f If you have problems check the path definitions locations of include files and libraries etc in the configure script and correct them if necessary Then re run configure to commit the changes to all makefiles the makemake script etc You can specify the user interface Cmdenv Tkenv with the u option with no u Tkenv is the default makemak u Tkenv The name of the output file is set with the o option the default is the name of the directory makemake o fddi net If some of your source files are generated from other files for example you use machine generated NED files write your make rules into a file called makefrag When you run makemake it will automatically insert makefrag into the resulting makefile With the i option you can also name other files to be included into makefile 7 2 3 Multi directory models In the case of a large project your source files may be spread across several directories You have to decide whether you want to use static linking shared or run time loaded libraries Here we discuss static linking In each subdirectory say trafgen and router run m
15. Two constructs stand out at once message and entity The message constructs define message types and are translated to C structs Entities correspond to OMNeT s simple modules and the entity body is equivalent to the activity function ename is a data type that holds entity references Restricted use of C Because the parser inside pcc is written for C it is not possible to use any C constructs This means you can NOT use e comments only e class libraries directly only wrapped into standard C functions Today when practically no one is programming in pure C this is a very serious limitation Again the limitation comes from pcc itself it doesn t understand any C It is irrelevant whether you use a C or C compiler to compile pcc s output Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix B Parsec and OMNeT 200 The driver entity assembling the model by hand The driver entity is special in a way it is similar to the C main function PARSEC starts the simulation by creating and running a driver entity The main task of the driver is to create all other entities in the simulation and provide them with information they need parameters values etc The latter is done by sending to the entities messages with the necessary parameters PARSEC does not have a high level topology description language like NED in OMNeT instead the driver entity is hand code
16. create redirection Now every operation you do on a s value will be done tobb long x a returns bb s value 10L a 5 bb s value changes to 5 The only way to determine whether a is really holding the value or it is redirected to another cPar is to use the isRedirected member function which returns a bool or redirection which returns the pointer to the background object or NULL if there s no redirection cPar redir a redirection returns bb s pointer if redir NULL ev lt lt a is redirected to lt lt redir gt name lt lt endl To break the link between the two objects use the cancelRedirection member function No other method will work including assigning a the value of another cPar object The cancelRedirection function gives the long 0 value to the redirected object the other will be unaffected If you want to cancel the indirection but keep the old value you can do something like this cPar value a redirection bb s pointer a cancelRedirection break the link value of a is now 0 a value copy the contents of bb into a 6 12 6 Type characters Internally cPar objects identify the types of the stored values by type characters The type character is returned by the t ype member function cPar par 10L char typechar par type returns L The full table of type characters is presented in the Summary section below TBD isNumeri
17. delete putAsideQueue remove msg else 6 4 7 Querying the state of an output gate You may have reasons to check whether a certain output gate is transmitting or to learn when it will finish transmission This is done with gate objects isBusy and transmissionFinishes member functions The latter function transmissionFinishes returns the time when the gate will finish its current transmission or if it is currently free when it finished its last transmission An example cMessage packet new cMessage DATA packet gt setLength 1000 if gate TxGate gt isBusy if gate is busy wait until it becomes free wait gate TxGate gt transmissionFinishes simTime send packet TxGate If the connection with a data rate is not immediately the one connected to the simple module s output gate but the second one in the route you have to check the second gate s busy condition You would use the following code if gate mygate gt toGate gt isBusy TEENA Note that if data rates change during the simulation the changes will affect only the messages that are sent after the change 6 4 8 Stopping the simulation Normal termination You can finish the simulation with the endSimulation function endSimulation However typically you dont need endSimulation because you can specify simulation time and CPU time limits in
18. double rnd histogram random cStdDev assumes normal distribution You can also wrap the distribution object in a cPar cPar rnd_par rnd_par rnd_par setDoubleValue amp histogram The cPar object stores the pointer to the histogram or P object and whenever it is asked for the value calls the histogram object s random function double rnd double rnd_par random number from the cPSquare Storing loading distributions The statistic classes have LoadFromFile member functions that read the histogram data from a text file If you need a custom distribution that cannot be written or it is inefficient as a C function you can describe it in histogram form stored in a text file and use a histogram object with loadFromFile You can also use saveToFile that writes out the distribution collected by the histogram object FILE f fopen histogram dat w histogram saveToFile f save the distribution fclose f FILE f2 fopen histogram dat r cDoubleHistogram hist2 Hist from file hist2 loadFromFile 2 load stored distribution fclose f2 Histogram with custom cells cVarHistogram class TBD comments Now we do support the following 2 uses of cVarHistogram e add all the boundaries manually before collecting samples e collect samples and transform makes the boundaries Transform types for cVarHistogram e HIST_TR_NO_
19. packet formats packet encapsulation There are no explicit packet formats in OMNeT However you can write function to create messages with specific fields and length EthernetFrame cMessage creat cMessage msg new cMessage msg gt setKind PACKET msg gt addPar source msg gt addPar destination msg gt addPar protocol msg gt setLength 8 16 return msg As in OPNET message parameters can be assigned object pointers thus also message pointers However there is also direct support encapsulation msg gt encapsulate innermsg innermsg msg gt encapsulatedMsg innermsg msg gt decapsulate ICIs are also represented by cMessage objects naturally with zero length If it is important to distinguish between packets and ICIs you can use the message kind field define PACK 0 define ICI 1 ET E Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix A OPNET and OMNeT ICI formats ICI attributes packet and ICI in the same interrupt cMessage pk new cMessage pk gt setKind PACKET cMessage ici new cMessage ici gt setKind ICI See packet formats See packet fields You can use encapsulation At the sender side cMessage ici pk ici gt encapsulate pk send ici out gate The receiver side 194 op_pk_send
20. BFS86 JCH85 PON91 PON92 PON93 KOF95 Jain Raj The Art of Computer Systems Performance Analysis Wiley New York 1991 Bratley P Fox B L and Schrage L E A Guide to Simulation Springer Verlag New York 1986 Jain Raj and Chlamtac Imrich The P Algorithm for Dynamic Calculation of Quantiles and Histograms without Storing Observations Communications of the ACM 28 10 1076 1085 1985 Pongor Gy rgy OMNET An Object Oriented Network Simulator 1991 Pongor Gy rgy Statistical Synchronization A Different Approach of Parallel Discrete Event Simulation Lappeenranta University of Technology Data Communications Laboratory Lappeenranta Finland 1992 Pongor Gy rgy On the Efficiency of the Statistical Synchronization Method European Simulation Symposium ESS 93 Delft The Netherlands Oct 25 28 1993 Kofoed Stig Portable Multitasking in C Dr Dobb s Journal November 1995 ftp ftp mv com pub ddj 1995 1995 11 mtask zip TBD include papers of Gabor Lencse OMNeT related research papers VAR99 VAR98a VAR98b V amp F97 V amp P97 Using the OMNeT Discrete Event Simulation System in Education Andras Varga IEEE Transactions on Education November 1999 CD ROM issue abstract in vol 42 no 4 pp 372 November 1999 K split On Line Density Estimation for Simulation Result Collection Andras Varga In the Proceedings of the European Simulati
21. Duplicates the object and returns a pointer to the new one cPar amp operator cPar amp otherpar The assignment operator works with cPar objects Redefined virtual functions virtual const char className Returns pointer to the class name string cPar Member functions handling type input flag and prompt char type Returns type character If the real type is I it returns the type of the object it is redirected to for example D L etc const char prompt Returns the prompt text or NULL void setPrompt const char s Sets the prompt text void setInput bool ip Sets ip t rue or clears p false the input flag bool isInput Returns t rue if the parameter is of input type the input flag is set bool changed Returns t rue if the value has changed since the last changed call Conversion from to text virtual bool setFromText const char text char type This function tries to interpret the argument text as a type typed value type means that the type is to be auto selected On success cPar is updated with the new value and true is returned otherwise the function returns false No error message is generated virtual void getAsText char buf int maxlen Places the value in text format it into buffer buf which is maxlen characters long Setting the value cPar amp setStringValue char tp const char s Sets the value stored in
22. File sample ned if This is a file comment File comments reach from the top of the file till the last blank line above the first code line The file comment can also contain blank lines so this is still part of the above file comment Modulel es This is a banner comment for the Modulel declaration below Banner comments can be multi line but they are not supposed to contain blank lines Otherwise the lines above the blank one will be taken as part of a file comment or trailing comment module Modulel submodules and this is right comment This is another banner comment for the submodule submodl Module display p 120 108 b 96 72 rect connections out gt submodl in Right comments can also be multi line endmodule Finally this is a trailing comment belonging to the above module It may contain blank lines Trailing comments are mostly used to put separator lines into the file like this Module2 Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 60 an empty module module Module2 endmodule Key mouse bindings In graphics view there are two editing modes draw and select mode The mouse bindings are the following Mouse Effect In draw mode Drag out a rectangle in empty
23. In derived classes these functions return the number of values collected the smallest largest value the mean the standard deviation the sum and the squared sum of the collected data respectively virtual double random The function generates a random number based on the collected data virtual void clearResult This function should be redefined in derived classes to clear the results collected so far oral virtual void saveToFile FILE Writes the contents of the object into a text file virtual void loadFromFile FILE Reads the object data from a file written out by saveToF ile or written by hand 15 12 2 class cStdDev cStdDev is designed to collect doubles and calculate data such as the minimum maximum value the mean and the standard deviation for them Construction destruction copying cStdDev cStdDev amp r cStdDev const char S NULL virtual cStdDev virtual cObject dup cStdDev amp operator cStdDev res Constructors destructor duplication and assignment Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 248 Redefined virtual functions virtual const char className j Returns a pointer to the class name string cStdDev virtual void info char buf virtual const char inspectorFactoryName virtual void writeContents ostream amp os Redef
24. Obtaining histogram data out of the object virtual void setRange double lower double upper virtual void setRangeAuto int num_firstvals double range_ext_fact virtual void setRangeAutoLower double upper int num_firstvals double range_ext_fact virtual void setRangeAutoUpper double lower int num_firstvals double range_ext_fact virtual void setNumFirstVals int num_firstvals Range setting virtual double pdf double x 0 virtual double cdf double x 0 Density function and cumulated density function at a given x virtual unsigned long underflowCell virtual unsigned long overflowCell Returns number of observations that fall out of the histogram range 15 12 5 Class cHistogramBase Base class for histogram classes 15 12 6 Class cVarHistogram Variable bin size histogram You may add cell bin boundaries manually or let the object create cells with equal number of observations in them or as close to that as possible Constructor cVarHistogram const char s NULL int numcells 11 int transformtype HIST_TR_AUTO_EPC_DBL The third argument can be one of HIST_TR_NO_TRANSFORM HIST_TR_AUTO_EPC_DBL HIST_TR_AUTO_EPC_INT New member function virtual void addBinBound double x If HIST_TR_NO_TRANSFOR M was passed in the constructor call you may specify cell bin bounds manually before collection starts
25. Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 86 Conventions for writing simple modules 1 Put the NED description the C class declaration and the implementation into three separate files Do not put two or more modules code into the same file unless they are build upon one another don t be afraid of small files Thus for a simple module called Foobar you should have Foobar ned Foobar h and Foobar cc This reduces coupling of module sources and makes your code more reusable Adopt a good coding style Some hints Choose your favourite indentation style and keep to that consistently I recommend four space indents and the brace placement style in which the OMNeT sources are written Write only one statement per line Avoid putting comments at the end of the line place them above the code on a separate line instead Use blank lines to break the code into not very long logical blocks and put a few word comment above each block what that block does Leave at least two blank lines between two member functions The purpose of all that is that the structure of your code be obvious at the first glance Identifiers Begin module type names with a capital letter and capitalize the beginning of each word like in TokenRingMAC Do not use underscore _ in module names Use the C style naming on member functions beginning of each word is capitalized except for the first one and
26. You can set a breakpoint inserting albreakpoint call into the source for cMessage msg receive breakpoint before processing Thies breakpoint before send send reply_msg out Tf vis In user interfaces that do not support debugging breakpoint calls are simply ignored 6 16 5 Disabling warnings Some container classes and functions suspend the simulation and issue warning messages in potentially bogus dangerous situations for example when an object is not found and NULL pointer reference is about to be returned Very often this is useful but sometimes it is more trouble You can turn warnings on off from the ini file warnings yes no It is a good practice to leave warnings enabled and temporarily disable warnings in places where OMNeT would normally issue warnings but you know the code is correct This is done in the following way bool w simulation warnings Simulation setWarnings false critical code Simulation setWarnings w Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 143 6 16 6 Getting coroutine stack usage If is important to choose the correct stack size for modules If the stack is too large it unnecessarily consumes memory if it is too small stack violation occurs From the Feb99 release OMNeT contains a mechanism that detects stack overflows It checks the intactness of a pr
27. dynamic module creation distributed execution tour of OMNeT samples using WATCH star topology with the number of modules as parameters distributed execution shows how to link several sim models into one executable Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 35 4 THE NED LANGUAGE 4 1 NED overview The description of model topology is given in the NED language The NED language supports modular description of a network This means that a network description consists of a number of component descriptions channels simple compound module types The channels simple modules and compound modules of one network description can be used in another network description As a consequence the NED language makes it possible for the user to build his own libraries of network descriptions Files containing network descriptions generally have a ned suffix Network descriptions are not used directly they are translated into C code by the NEDC compiler then compiled by the C compiler and linked into the simulation executable The EBNF description of the language can be found in the appendix 4 1 1 Components of a NED description A NED description can contain the following components in arbitrary number or order e import statements e channel definitions e simple and compound module declarations e system module declarations The rest of this chapter disc
28. gt has been sent using the send message to entity after x clause gt gt Is this possible in PARSEC No gt I d like to do this for activating and cancelling timers here are other ways to implement timers in Parsec particularly f the timers are sent to self One way is to keep a list of imeout values and set the X in timeout in X to the minimum in he list i e use timeout don t send messages To cancel imply remove from the list Another way is to send to self but eep a list of cancelled messages in a circular queue for example hen the message is received check to see if it was cancelled and if so discard it S wun tteey Rich Simulation clock The PARSEC simulation clock is of integer type optionally unsigned int or long long long long is not a standard ANSI C C type OMNeT uses double It is probably application specific which is the better choice but in the case of a large simulation model put together from components written with different time granularity in mind double seems a better choice because it is relative insensitive to the choice of the time unit Thread coroutine handling Although the Parsec documentation says nothing symbol names is the Parsec runtime library give the impression that the thread coroutine implementation is quite similar in OMNeT and the single processor implementation of Parsec On Unix systems bo
29. module Router parameters rte_processing_delay rte_buffersize num_of_ports const gates submodules connections endmodule Compound module parameters can be used in two ways e used in expressions for submodule parameter values e used in defining the internal topology of the network For example a parameter called num_of_ports can be used to construct a router module with the number of ports as a parameter Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 39 4 5 2 Compound module gates Gates have the same role and are declared in the same way as with simple modules Please refer to the Simple module gates section Example module Router parameters gates in input_port out output_port submodules connections endmodule 4 5 3 Submodules Submodules are defined in the submodules section of a module description For each submodule there are sections to define the actual values to be passed to its parameters and the sizes of its gate vectors Example module NameForCompoundModule parameters gates submodules SubModuleName TypeOfSubModule parameters 2 gee gatesizes ees SecondSubModuleName TypeOfSecondSubModule a oe connections endmodule In a submodule definition one has to supply the name of a previously defined module as the type and a module name The description of the module
30. 10 PARALLEL EXECUTION 5 ss3sosihsescaiecesaveasssosbaxcocs3 cossbovsiaccssssocdsseansoaassvesssoossavonstaess 175 11 THE DESIGN OF OMNET sssscisesssscescseasssoossesessessnsoscesecnstnehoassgenseesasisetensossssounaties 180 12 APPENDIX A OPNET AND OMNET ccccsccsscccccscscccssccccssccscscsccsscssceeseses 186 13 APPENDIX B PARSEC AND OMNET esssseeesoosessooesoosessooecooesesosecsoesesoosesoeeesoosess 198 14 APPENDIX C NED LANGUAGE GRAMMAR ssesssocessoesssoccesoeessoocesoesesoceesseeesoosesse 206 15 APPENDIX D CLASS LIBRARY REFERENCE essssocessoeessocessoesesoocesoeeesooeesoseeesoosesse 210 16 REFERENCES oissocddssessdvensscnedecatsssnessosnesgusnsoehaseesandocssesensssesseh cuseceahsouseassucstennssensissnes 257 Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 3 CONTENTS 1 INTRODUCTION aio vacegs odseseesaucesscesccasssensstiesty ceccaabcsscassevazanevseuaucessoucensteatnestiqaventeseciaesees 14 11 What is OMNeTHE 7 vcsssseasaac diviieis essed Sead ean elena aden ital nea ao ede 14 1 2 Where is OMNeT in the world of simulation tools 00 0 eeeeeceeeeeeeteeeeeteeeeeteeeeneeeenes 15 13 Organization of this manual sissies a a E ans 16 EES Cn pee re E E O E EE E ES T E E E 17 1 5 Authors esas iE sa sua EE ERE ETE AE AO e eA 17 Die OVERVIEW siccssccteccsssctsacesueestastvodeaietcstisscen odsduesanteanee E E a e a s eias 19 2 1 Modeling CONCEPES ensercusurnn a na E E
31. Dept of Telecommunications OMNeT Manual The Simulation Library token num_messages omitted cModulePar token ia_time begin Type F Value truncnormal 0 005 0 003 end rest of parameters amp gates stuff deleted from here cCompoundModule token comp 0 begin parameters cArray empty gates cArray n 2 mac TokenRingMAC 3 gen Generator 4 sink Sink 5 end cArray token comp 0 parameters begin end cArray token comp 0 gates begin in cGate lt comp 2 out out cGate gt D gt comp 1 in end cGate token comp 0 in begin type input inside connection token comp 0 mac phy_in outside connection token comp 2 out delay error data rate end cGate token comp 0 out begin type output inside connection token comp 0 mac phy_out outside connection token comp 1 in delay cPar le 0O6 D error data rate end TokenRingMAC token comp 0 mac begin parameters cArray n 2 gates cArray n 4 local objects cHead class data members cHead putaside queue cQueue empty end comp 0 mac parameters stuff deleted from here cArray token comp 0 mac gates begin phy_in cGate lt lt parent gt in from_gen cGate lt gen out phy_out cGate gt lt parent gt out Technical University of Budapest Dept of Telecommunications 140 OMNeT Manual The Simu
32. FES to order messages with identical arrival time values void setLength long 1 Sets message length When the message is transmitted through a channel its error flag will be set with a probability depending on the length of the message and the channel s bit error rate void setError bool err Directly sets the message s error flag void setTimestamp Sets the message s time stamp to the current simulation time void setTimestamp simtime_t stamp Directly sets the message s time stamp int kind Returns message kind The message kind member is not used by OMNeT 4 it can be used freely by the user Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 226 int priority Returns message priority The priority member is not used by OMNeT it can be used freely by the user long length Returns message length bool hasBitError Returns t rue if error flag is set false otherwise simtime_t timestamp Returns the message s time stamp Message encapsulation void encapsulate cMessage msg Encapsulates msg in the message msg gt length will be added to the length of the message cMessage decapsulate Decapsulates a message from the message object The length of the message will be decreased accordingly except if it was zero If the length would become negative an error occurs cMessage encapsulatedMsg
33. In this case when the object s value is first used the parameter value will be searched for in the configuration ini file if it is not found there the user will be given a chance to enter the value interactively Examples cPar inp inp inp setPrompt Enter my value inp setInput true make it an input parameter double a double inp the user will be prompted HERE 6 12 2 Random number generation through cPar Setting cPar to call a function with constant arguments can be used to make cPar return random variables of different distributions cPar rnd rnd rnd setDoubleValue intuniform 10 0 10 0 uniform distr rnd setDoubleValue normal 100 0 5 0 normal distr mean dev rnd setDoubleValue exponential 10 0 exponential distr mean intuniform normal etc are ordinary C functions taking double arguments and returning double Each time you read the value of a cPar containing a function like above the function will be called with the given constant arguments e g normal 100 0 5 0 and its return value used The above functions use number 0 from the several random number generators To use another generator use the genk__xxx versions of the random functions rnd setDoubleValue genk_normal 3 100 0 5 0 uses generator 3 A cPar object can also be set to return a random variable from a distribution collected by a statistical data co
34. Manual Analyzing Simulation Results 172 files on disk In the right pane you can duplicate vectors if you want to filter the vector and also keep the original If you set the right options for a vector but temporarily do not want it to hang around in the right pane you can put it back into the left pane for storage 9 1 3 Writing filters Filters get an output vector on their standard input as plain text with the timestamp being the second and the value being the third field on each line do some processing to it and write the result to the standard output Filters can be incorporated in one of three ways as awk expressions as awk programs or as external programs An awk expression filter means assembling and launching a command like this cat foobar vec awk 3 lt expression gt print An awk program filter means running the following command cat foobar vec awk lt program gt The third type of filters is used like this cat foobar vec lt program gt lt parameters gt Before the filter pipeline is launched the following substitutions are performed on the awk scripts t gt 2 x gt 3 The parameters of the form paramname are also replaced with their actual value Thus if you want to add 1 to all value you can use the awk expression filter x 1 It will turn into awk 3 3 1 print When you want to shift the vector by a used defined DT time you can creat
35. cArray token comp 0 mac local objects send queue 0 gt 1l par vector begin dest cPar 1 L source cPar O L gentime cPar 0 0158106 D end message parameters and the other messages stuff deleted cHead token comp 0 mac class data members begin end cQueue token comp 0 mac putaside queue begin end comp 0 gen and comp 0 sink stuff deleted from here Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 142 whole comp 1 and comp 2 stuff deleted from here cMessageHeap simulation message queue begin 1 gt 0 cMessage Tarr 0 0576872457 57ms Src 8 Dest 7 cMessage Tarr 0 0577201630 57ms Mod 8 selfmsg cMessage Tarr 0 0585677054 58ms Mod 4 selfmsg cMessage Tarr 0 0594939072 59ms Mod 12 selfmsg cMessage Tarr 0 0601010000 60ms Mod 7 selfmsg 1 gt 2 cMessage Tarr 0 0601020000 60ms Src 11 Dest 13 end detailed list of message queue contents deleted from here To reduce the size of the file you may well decide to make a snapshot only of a part of the model This example reports only about the current simple module s put aside queue snapshot amp putAsideQueue 6 16 4 Breakpoints With activity only In those user interfaces which support debugging breakpoints stop execution and the state of the simulation can be examined
36. connections endmodule Host names propagate up to network definition level Extension to the network definition network distVector DistVector on machinel machine2 machine3 endnetwork The on parameters of the network definition can be actual host names or alternatively they can be symbolic names that are mapped to actual host names in the config file 4 8 2 Conditional on sections Similarly to the parameters and gatesizes section multiple on sections can exist for the submodules if they are tagged with if phrases This makes it possible to control the module distribution with parameters You can even put different parts of a module vector on different machines using the index operator see later in the section describing expressions Example module DistVector machines hostl1 host2 host3 submodules node Node count on if index lt count 33 hostl on if index gt count 33 amp amp index lt count 66 host2 on if index gt count 66 host3 endmodule network distvector DistVector on machinel machine2 machine3 endnetwork Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 53 4 9 Expressions In the NED language there are a number of places where expressions are expected When such an expression is encountered by the NEDC compiler it is compiled and it will be evaluated run time Expressions have a C style syntax They are built
37. const char s The same as get int get const char With the indexing operator cArray can be used as a vector bool exist int m bool exist const char s Returns t rue if the value returned by get int get const char would not be null reference NOOBJ cObject remove int m cObject remove const char s cObject remove cObject p Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 236 Removes the object given with its index name pointer from the container If the object was owned by the container drop is called 15 9 2 class cQueue cQueue is a container class that can hold objects derived from cOb ject cQueue acts as a priority queue The user must provide a function that can compare two objects If no such function is given cQueue implements a FIFO Order ascending or descending can be specified and will be interpreted as in the figure ttt insertion removal Ownership of contained objects responsibility of deletion can be specified per object basis see cObject takeOwnership Default is that cQueue takes the ownership of each object inserted that is takeOwnership true The sorting function int CompareFunc cObject amp a cObject b User supplied sorting functions must have a declaration like the one shown above and they must return a negative value if a lt b 0 if a b and a positive value if a gt
38. event will occur is the message s destination module and the model time when the event occurs is the arrival time of the message Events like timeout expired are implemented with the module sending a message to itself Simulation time in OMNeT is stored in the C type simtime_t which is a typedef for double Events are consumed from the FES in arrival time order to maintain causality More precisely given two messages the following rules apply 1 the message with earlier arrival time is executed first If arrival times are equal 2 the one with smaller priority value is executed first If priorities are the same 3 the one scheduled or sent earlier is executed first Priority is a user assigned integer attribute of messages Storing simulation time in doubles may sometimes cause inconveniences Due to finite machine precision two doubles calculated in two different ways do not always compare equal even if they theoretically should be This means that if you want to explicitly rely on the arrival times of two events being the same you should take care that simulation times which should be equal are calculated in exactly the same way Another possible approach is to avoid equal arrival times for example by adding subtracting small values to schedule times to ensure specific execution order inorder_epsilon We also thought about some simtime_precision parameter in the simulation kernel that would force t and t2 to be regarde
39. gate ordered by time Simulation executes normally until it comes to an event that has a time definitely past the first syncpoint in the list That event is not processed but the segment goes into a blocked state While the segment is blocked it listens for messages arriving from other segments In the actual implementation passive wait is used so a blocked segment doesn t use much CPU time Each message that arrives deletes the first syncpoint in the list that matches its gate The segment goes out of the blocked state when because of deletions the first syncpoint in the list is no longer past the event in question Then the simulation goes on normally either with the newly arrived message or the earliest of them or the original event A message that arrives outside of the blocked state also causes deletion of the first matching syncpoint in the list this case corresponds to the reverse case when the sender segment is ahead of the receiving segment in model time Deadlock It is possible to cause deadlock with carelessly placed syncpoints Suppose that segment A declares a syncpoint at 10s with segment B but it will actually send a message only at 10 5s If segment B does the same to segment A a nice deadlock is created OMNeT makes no effort to detect or prevent such deadlocks it is entirely the simulation programmer s task to take care that deadlocks do not occur 10 1 3 The PVM virtual machine The pvmhosts file is used
40. i gate name gate gt toGate fromGate gate gt destinationGate sourceGate gate gt ownerModule op_pro_create See dynamic module creation Note that this is a more powerful tool than OPNET s dynamic processes in that you can also create compound modules Prohandle Module ID Given the module pointer you can obtain module ID by int id mod gt id And you can obtain module pointer from the ID cModule mod simulation module id An invalid ID is negative op_pro_invoke op_pro_destroy op_pro_destroy self Dynamically created modules do not need to be invoked they live their own life To dispatch messages to them you can use sendDirect deleteModule module deleteModule Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix A OPNET and OMNeT 196 module memory parent to child memory argument memory to dynamic processes Parent module can set pointers void data members in the dynamically created module object any time thus also right after creating it gt parent to child memory right before sending a packet to it gt argument memory and the pointer can refer to memory managed by the parent module gt module memory An example for argument memory Suppose the child module class has a public data member named argmem class ChildModule
41. member function and because cPar supports typecasting and assignment it is easy to read and set the value of a parameter long dest_addr msg gt par dest_addr msg gt par dest_addr 168 The addPar function also returns a reference to the added cPar object so you can set the value of the new parameter at the same place msg gt addPar dest_addr 168 You can use the hasPar function to see if the message has a given parameter or not Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 98 if msg gt hasPar dest_addr msg gt addPar dest_addr Numeric indices Message parameters can be accessed also by index in the parameter array The findPar function returns the index of a parameter or 1 if the parameter cannot be found The parameter can then be accessed using an overloaded par function Access by index is more efficient than access by name although access by name might become faster in the future by using hashtables long dest_addr 0 int index msg gt findPar dest_addr if index gt 0 dest_addr msg gt par index Adding arbitrary data by accessing the internal array Message parameters are stored in an object of type cArray which can store any object type not only cPars The parList member function lets you directly access the internal cArray so by calling cArray s member functions you can attach any object
42. node j from_downright if rightchild i j node i from_up lt node j downright if rightchild i j endfor Thee s endmodule The dotted lines should be replaced by modules that close the tree at its root and the lower edge The leftchild i j and rightchild i j functions are leftchild i j 1 if i 2j 1 0 otherwise rightchild i j 1 if i 2j 2 0 otherwise These formulas can be directly substituted in the NED description or alternatively written in C and linked into the simulation executable Example 4 Random graph Conditional connections can also be used to generate random topologies The following code generates a random subgraph of a full graph module RandomGraph parameters count const connectedness 0 0 lt x lt 1 0 submodules node Node count gatesizes in count out count connections nocheck for i 0 count 1l j 0 count 1 do node i out j gt node j in i if i j and uniform 0 1 lt connectedness endfor endmodule Note that not each gate of the modules will be connected By default an unconnected gate produces a run time error message when the simulation is started but this error message is turned off here with the nocheck modifier Consequently it is the simple modules responsibility not to send on a gate which is not leading anywhere 4 6 2 Using const with parameterized topologies Since parameter values can be used in defining the internal topology of the module the const mo
43. par pointerValue par setValue void 0O gt now par has nothing to do with the double 15 array However if you assign some non pointer value to the cPar beware this will activate the memory management for the block If you temporarily use the same cPar object to store other than void P values the configPointer setting is lost 6 12 4 Reverse Polish expressions This feature is rarely needed by the user it is more used internally A cPar object can also store expressions In this case the expression must be given in reversed Polish form An example sXElem expression new sXElem 5 Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 124 expression 0 amp par count pointer to module parameter expression 1l 1 expression 2 expression 3 2 expression 4 cPar expr expr expr setDoubleValue expression 5 The cPar object created above contains the count 1 2 expression where count is a module parameter Each time the cPar is evaluated it recalculates the expression using the current value of count Note the amp sign in front of par count expression if it was not there the parameter would be taken by value evaluated once and then the resulting constant would be used Another example is a distribution with mean and standard deviation given by module parameters sXElem expressio
44. pointers Used by cLinkedList and cPar Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 211 15 2 3 Simulation time conversion functions char simtimeToStr simtime_t t char dest NULL Converts simtime_t a double to textual form that contains hours minutes second milliseconds and microseconds like this OOh 34m 23s 130ms 230us If you do not provide a destination buffer for simt imeToStr it will place the result into a static buffer that is overwritten with each call simtime_t strToSimtime const char str Converts a textually given simulation time e g 30s 152ms to simtime_t If the string cannot be entirely interpreted 1 is returned simtime_t strToSimtimeO const char amp str Similar to st rToSimt ime it can be used if the time string is a substring in a larger string It only goes as far as it can in str then sets the st r pointer to the first character that could not be interpreted as part of the time string and returns the value It never returns 1 if nothing at the beginning of the string looked like simulation time it returns 0 15 2 4 Value added string functions These functions replace some of the lt string h gt functions The difference is that they also accept NULL pointers and treat them as pointers to an empty string and use operator new instead of malloc Use these functions instead of original lt string
45. section of a module description An empty bracket pair denotes a gate vector Elements of the vector are numbered from zero Examples simple DataLink parameters gates in from_port from_higher_layer out to_port to_higher_layer endsimple simple RoutingModule parameters gates in output out input Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 38 endsimple The sizes of gate vectors are given later when the module is used as a building block of a compound module type Thus every instance of the module can have gate vectors of different sizes 4 5 Compound module definitions Compound modules are modules that are composed of one or more submodules Compound modules like a simple modules can have parameters and gates so a compound module definition looks similar to a simple module definition except that it also has sections to specify the submodules and connections within the module Submodules can either be simple or compound modules they are equivalent Example module SomeNameForCompoundModule parameters ieee gates bid des submodules PES ts connections fie ee endmodule Any of the above sections parameters gates submodules connections is optional 4 5 1 Compound module parameters Parameters are declared in the same way as with simple modules Please refer to the Simple module parameters section Example
46. this is done by discarding negative values until a nonnegative comes 15 2 7 Miscellaneous Min a b Max a b Macros returning the minimum maximum of two values bool equal double a double b double limit Tests equality of two doubles with the given precision Returns t rue if fabs a b lt limit bool lowmemory Returns t rue if free space on the heap is getting too low ASSERT condition Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 213 Makes sure the condition holds evaluates to nonzero Otherwise it stops the simulation with a detailed error message Use ASSERT to verify your program works correctly and what you assume about program state is really true 15 2 8 String class class opp_string A value added version of char Has only one data member a char pointer Added value is automatic allocation deallocation through opp_strdup delete A string object has its own opp_strdup ped copy of the string Recommended use as class member where otherwise the class members would have to call opp_strdup and delete for the char member It is intentionally kept very simple Usable wherever char is needed Its only data member is the char pointer so st ring is even usable with printf which doesnt do conversion to char To keep this very valuable property don t add any other data members or any virtual functions Virtual
47. thus parameter 0 2 normal 100 5 mean 100 variance 5 truncnormal 5 3 normal distr truncated to nonnegative values The functions all use the random number generator 0 By using the genk_ prefixed versions of the above functions you can specify which generator should be used The index of the generator comes as the first argument Example genk_normal 2 100 5 as normal 100 5 using generator 2 The above distributions are implemented with C functions see later in the Functions section This also means that you can easily add further ones by writing their code in C and using the Register _Function macro Your distributions will be treated in the same way as the built in ones 4 9 6 Input value The syntax is input 10 Number of processors Or you can omit the prompt text input 10ms Value for input parameters can be given in the config file If they are not there the user will be offered a prompt to enter the value 4 9 7 Functions In NED expressions you can use mathematical functions e many of the C language s lt math h gt library functions exp log sin cos floor ceil etc e functions that generate random variables uniform exponential normal and others were already discussed e user defined functions that can implement new functions or yield random variables of distributions that are originally not built in To use user defined functions one has
48. which should do the allocation and the appropriate copying Deletion is done by calling the user supplied delete function or the delete operator if it is not supplied Contained items will be deleted as configured by Returns the number of items contained in the list bool empty Returns t rue if the list is empty void insert void p Inserts the given object into the list maintaining the sorting order void insertBefore void where void insertAfter void where void p void p Inserts exactly before and after the given item void tail Returns the last item in the list or null pointer if the list is empty void pop Unlinks and returns the last item in the list void remove void obj Unlinks and returns the given item Technical University of Budapest Dept of Telecommunications size_t size OMNeT Manual Appendix D Class Library Reference 241 15 10 2 class cLinkedListiterator To examine each element in the list the cLinkedListIterator iterator class can be used Once acLinkedListIterator object is created for the list the cLinkedList object the and operators can be used to step from one element of the list to the next previous one cLinkedListiIterator is not a descendant of cObject or cIterator cLinkedListIterator cLinkedList amp ll int a 1 Constructor cIterator will walk on the list passed as argument The current item
49. 1 The central object cSimulation simulation The cSimulation class stores a network and manages simulation There is only one instance a global object called simulation The object has two basic roles e asa vector of modules e holds global variables for example the message queue 11 3 2 Module classes Base class for module classes cModule Two derived classes cCompoundModule cSimpleModule User simple modules are derived from cSimpleModule A cModule has array of parameters array of gates member functions to set up and query parameters and gates cSimpleModule adds put aside queue list of local objects the virtual function activity member functions like send receive etc Gates are represented by the cGate objects Connections are not real objects their attributes delay error datarate are managed by the connection s source gate 11 3 3 Global registration lists There are global objects holding lists of components available in an OMNeT executable These lists are List object Macro that creates a Function member Class of members Technical University of Budapest Dept of Telecommunications OMNeT Manual The Design of OMNeT 182 cHead networks cHead modtypes Define_Network cNetworkType Define _Module Define_ Module _Like cModuleType List of available networks A cNetworkType object holds a pointer to a function that can build up the network
50. Deal Simple mod le libraries inen a e E E 86 5 7 2 Compound module NED source libraries sssseesseeessseesseesseessessseeesseeessresseesse 87 5 7 3 Precompiled compound module libraries ceeeeeseceeseeceeeeececsteeeenteeeenseeeeees 88 5 8 SOME simulation LECOTIGUES lt s 204 cas zuegheanseeysteocuieaseteaactauadaesy e E a ea EES EEr 89 5 8 1 Modeling computer networks esseeessseesssessesssessseessseesstessetsseesseeesseeesseesseesse 89 5 8 2 Modeling multiprocessor systems sssssssessesesessseeessetessessetsseesseeesseeesseesseesse 89 3 8 3 Parameter CIID sreci ens e o E aia a Eea 90 5 8 4 Multiple experiments within one simulation run ssssesssessssssessseesssseessresseesse 90 Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 6 5 8 5 Dynamic topology OptiMIZatiONn ce eeeeeececsscecssececseeeeceeeecseeecseeeeseeeenaeeeenas 91 6 THE SIMULATION LIBRARY iss ccsstencssacecescacstasadscescatvsaienccssacacashadsbbdatecusestusatenacseaseosre 92 621 Class library CONV CMU OMS sisses ee e a bens eae at i sad eei rge 92 62 WIGS safc Beaute e a ae a a e a a a a TA 95 6 3 Messages and Packets sai neresnic eoig ein r EEE a E EE EE ER 96 6 3 1 The cMe ssage Classiciin torpant a En aE E EEE A E Oai EEREN 96 6 3 2 Attaching parameters and objects to a message sssssesssessssssersseesssseessressers 97 6 3 3 Message encapsulation ssssssss
51. However there is a more straightforward method than the above two and this is delayed sending Delayed sending can be done with one of these functions sendDelayed cMessage msg double delay const char gate_name int index sendDelayed cMessage msg double delay int gate_id The arguments are the same as for send except for the extra delay parameter The effect of the function is the same as if the module had kept the message for the delay interval and sent it Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 102 afterwards That is the sending time of the message will be the current simulation time time at the sendDelayed call plus the delay The delay value must be nonnegative Example sendDelayed new cMessage token 0 005 out gate 6 4 3 Direct message sending Sometimes it is necessary or convenient to ignore gates connections and send a message directly to a remote destination module The sendDirect function does that and it takes the pointer of the remote module cCModule You can also specify a delay and an input gate of the destination module cModule destinationmodule double delay truncnormal 0 005 0 0001 sendDirect new cMessage delay destinationmodule in The destination module receives the message as if it was sent normally 6 4 4 Receiving messages With activity only T
52. If the parameter was taken by reference with a ref modifier in the NED file other modules will also see the change Thus parameters taken by reference can be used as a means of module communication An example par wait_time 0 12 Or cPar amp wait_time par wait_time wait_time 0 12 See cPar explanation later in this manual for further information on how to change a cPar s value 6 5 2 Gates and links Gate objects Module gates are cGate objects Gate objects know whether and to which gate they are connected and they can be asked about the parameters of the link delay data rate etc The gate member function of cModule returns a pointer to a cGate object and an overloaded form of the function lets you to access elements of a vector gate Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 107 cGate outgate gate out cGate outvecdSgate gate outvec 5 For gate vectors the first form returns the first gate in the vector at index 0 The isVector member function can be used to determine if a gate belongs to a gate vector or not But this is almost insignificant because non vector gates are treated as vectors with size 1 Given a gate pointer you can use the size and index member functions of cGate to determine the size of the gate vector and the index of the gate within the vector size2 outvecdSgate gt size gt si
53. NED and config files e g 10h 30m 45 25 As with parameter names wildcards are allowed in the object names and module path names An example Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 159 omnetpp ini OutVectors interval 1s 60s End to End Delay enabled yes Router2 enabled yes enabled no The above configuration limits collection of all output vectors to the 1s 60s interval and disables collection of output vectors except all end to end delays and the ones in any module called Router2 8 2 5 Module parameter logging It is possible to log all the changes to module parameters into a text file This can be useful when the simulation contains run time tuning of one or more module parameters and one wants to have the trajectory documented Module parameter logging must be explicitly enabled from the header file if one wants to use it General log parchanges yes parchange file token pch The format of the parameter change file is similar to the that of the output vector file If a parameter is taken by reference by several modules any change to the parameter will appear in the file under the name of the top level parameter no matter which module actually changed it and under what name 8 2 6 Display strings Display strings control the modules graphical appearance in the Tkenv user interface Displ
54. Returns a pointer to the encapsulated message or NULL Info about sending scheduling int senderModuleld Returns sender module s index in the module vector or 1 if the message hasnt been sent scheduled yet int senderGatelId Returns index of gate sent through in the sender module or 1 if the message hasnt been sent scheduled yet int arrivalModuleld Returns receiver module s index in the module vector or 1 if the message hasnt been sent scheduled yet int arrivalGatelId Returns index of gate the message arrived on in the sender module or 1 if the message hasn t sent scheduled yet cGate senderGate cGate arrivalGate Returns pointers to the gate from which the message was sent and on which gate it arrived A NULL pointer is returned for new unsent messages and messages sent viascheduleAt simtime_t creationTime Returns time when the message was created simtime_t sendingTime Returns time when the message was sent scheduled or 0 if the message hasn t been sent yet simtime_t arrivalTime Returns time when the message has arrived or 0 if the message hasn t been sent scheduled yet bool arrivedOn int g Return t rue if the message has arrived through gate g Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 227 bool arrivedOn const char s int g 0 Return t rue if the message has arrived throu
55. Simple modules in COMING Tatas 32s yasseieiae vas oacah naatcooahi es Goadows aces ents 62 JAA Events in OMNe PHS isco se seas a gee ey ceeded as Sea Cased od sal a Goats Sal ae ue 63 Dl Dy EESamplementat ON sesh Fs ial sco ics sansa eas Bek wee ace alec Dee 63 5 2 Defining simple module types cids sesicscesseccisdscaciaspasadsa sousaccennceads sdesedeasoesaceesedenessaeecsaneseneee 64 Bide OyeryieW sirsie ae E pa E a as ted Edel update daadl iiia 64 5 2 2 The module declaration sjsu ved patria uae edie ieee 64 5 2 3 Several modules single NED interface eeeccecesccecsseceeseececeeeeeseeeenseeeees 66 5 2 4 The class declaration vss sins ceili elena ld BANG ele ele 66 5 2 5 Decomposing activity handleMessage and inheritance eeeeseeeeeees 68 5 3 Adding functionality to cSimpleModule eeeceeceeseccesseeconteccesseccesneneensecenseccensenes 69 Dds We ACULVELY O enaa wecoeecaetaws 8s We cans rach tus coad fn caluds N M us coat once thas 69 5 3 2 NAMIE ME SSAC sees nce decti ce vais lt ans cee sacy ous ncaa ecb asu cu A eos A i 713 5 3 3 imitialize and Fish eesin a a nania 78 5 4 Finite State Machines in OMNeT H essesesessesesesresseesrerressersresressreseeseresresseseresresseseresees 80 5 5 Message transmission modeling eeeeesssessesessessresresstesrerresseesttseessresteseresresseseresresseseresees 84 5 6 Coding CONVENTIONS seei nhe n aee EE EE R E a EEE EEE e A ias 85 J Re OUP OMS HL BEAL ES a a Aa 86
56. Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 157 If set for all runs together If set for individual runs settings for slave processes general settings General Run 1 Run 2 etc user interface specific settings Cmdenv Tkenv etc Run 1 Run 2 etc module parameter values Parameters Run 1 Run 2 etc output vector configuration OutVectors Run 1 Run 2 etc graphical appearance DisplayStrings Run 1 Run 2 etc logical physical machine Machines not possible mappings with distributed execution Slaves Run 1 Run 2 etc The most important options of the General section are the following The ini warnings option can be used for debugging ini files if enabled it lists which options were searched for but not found The network option selects the model to be set up and run The length of the simulation can be set with the sim time limit and the cpu time limit options the usual time units such as ms s m h etc can be used The warnings option enables disables run time warnings it is recommended to be turned on while debugging The distributed flag selects between normal and parallel execution The output file names can be set with the following options snapshot file output vector file output scalar file parchange file For the last file
57. and conversion for fundamental data types TBD update cPar amp operator char c Equivalent to setValue C int c cPar amp operator unsigned char c Equivalent to setValue C int c cPar amp operator const char s Equivalent to setValue S s cPar amp operator int i Equivalent to setValue L long i cPar amp operator unsigned int i Equivalent to setValue L long i cPar amp operator long 1 Equivalent to setValue L 1 cPar amp operator unsigned long 1 Equivalent to setValue L long 1 cPar amp operator double d Equivalent to setValue D d cPar amp operator long double d Equivalent to setValue D double d cPar amp operator cObject obj Equivalent to set Value P obj operator char Equivalent to stringValue operator int Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 233 Equivalent to longValue operator unsigned int Equivalent to longValue operator long Equivalent to longValue operator unsigned long Equivalent to LlongValue operator double Equivalent to doubleValue operator long double Equivalent to doubleValue operator cObject Equivalent to pointerValue 15 8 2 Overloaded arithmetic operators f
58. and the way the internal connections are made Compound modules can pass parameters or expressions of parameters to their submodules Parameter passing can be done by value or by reference During simulation execution if a module changes the value of a parameter taken by reference the changed value propagates to other modules This effect can be used to tune the model or as a second means of module communication Pointer valued parameters can be used to implement shared memory 2 1 6 Topology description method The user defines the structure of the model in NED language descriptions NEtwork Description The NED language will be discussed in detail in Chapter 4 Technical University of Budapest Dept of Telecommunications OMNeT Manual Overview 23 2 2 Programming the algorithms The simple modules of a model contain the algorithms as C functions The full flexibility and power of the programming language can be used supported by the OMNeT simulation class library OMNeT supports a process style description method for describing activities During simulation execution simple module functions appear to run in parallel because they are implemented as coroutines also termed lightweight processes Coroutines were chosen because they allow an intuitive description of the algorithm and they can also serve as a good basis for implementing other description methods like state transition diagrams or Petri nets OMNeT has a co
59. available in a single linked list allows OMNeT to instantiate module types given only their class names as strings without having to include the class declaration into any other C source The global object also stores the name of the NED interface associated with the module class The interface description object another object generated by nedc is looked up automatically at network construction time Whenever a module of the given type is created it will automatically have the parameters and gates specified in the associated interface description Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 66 5 2 3 Several modules single NED interface Suppose you have three different C module classes TokenRing_MAC Ethernet_MAC FDDI_MAC which have identical gates and parameters Then you can create a single NED declaration General_MAC for them and write the following module declarations in the C code Define_Module_Like TokenRing_MAC General_MAC Define _Module_Like Ethernet_MAC General_MAC Define _Module_Like FDDI_MAC General_MAC In this case you wont be able to directly refer to the TokenRing_MAC Ethernet_MAC FDDI_MAC module types in your NED files For example you cannot write module PC submodules mac Ethernet_MAC error Ethernet_MAC not defined endmodule However you can pass the module type in a string valued param
60. c2 8 elo cilied edie ie Atel aeacetidn lan ats 225 15 6 1 Message base class cMessage ssenssessseeeesseesseesseessereseessseessresseesseessseeessees 225 15 6 2 Network packet class cPacket o cionceneen e S 227 15 7 Module sate class Gate iac isy saci sesceastansdaazsinnina ca dessaacasuscaaetandodecasaadan A E aa i 227 15 7 1 classveGate ntsi ernis ia eae etree ee eee eae aed 228 15 95 Value holding Class CR at cca 2a 3i2s wis coeecaecaus 8 Mogesea teractusaasacades egies dae aceuatevcedaaiarecaetsanenens 229 PS 8 Class CP atin EE ae eae ote ia ees cade eae 230 15 8 2 Overloaded arithmetic operators for CPar e ee eeeeeeeeeeeeseeeeceteeeceeeeeesneeeenaeeees 233 158 3 Sttuchsx Blem 32 6 ne Gh Be Oe OE ON eine BE AN 233 15 9 Object container classes cArray cQueue CHead eee eeeececeeececssececeeeeeceteeeeseeeeeeaeeees 234 15 9 CLASS CAT AY koirat irte tose lesnatayatas te a av e ta e e a o a 234 E V NO A E E E A E E E E E ES 236 159 3 class cQ eyelteralor annen oa a a E Ei 237 15 9 4 IS TUNC ONS reren in a a a E E 237 159 5 CLASS CCA ah i nes ee ae a e es ate cee ae orsta e te aas 238 Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 12 15 9 6 str ct CHIERA COT seini a uastanhu cia iis dma A E E 239 15 10 Non object container classes cLinkedList CBag cceeecccesscecesececeeeeeceteceenteeeenaeeees 239 PAO sT class C bined tS hai ct cealaee e ee ailel os a cael i an S
61. can be written in a releatively short time and it is inexpensive to maintain afterwards if the structure of the data files change the NED creating program can be easily modified The disadvantage is that the resulting NED files are often quite big and the C compilation of the _n cc files take too long Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 146 This method is best suited in the first phase of a simulation project when the topology the format of the data files etc have not yet settled down 6 19 2 Building the network from C code Another alternative is to write C code which becomes part of the simulation executable The code would read the topology data from data files or a database and build the network directly The code which builds the network would be quite similar to the _n cc files output by nedc Since writing such code is more complex than letting perl generate NED files this method is recommended when the simulation program has to be somewhat more productized for example when OMNeT and the simulation model is embedded into a larger program e g a network design tool Technical University of Budapest Dept of Telecommunications OMNeT Manual Building Simulation Programs 147 7 BUILDING SIMULATION PROGRAMS 7 1 Overview As it was already mentioned an OMNeT model physically consists of the following parts NED language topology descript
62. can exploit the fact that each node is connected to N others which differ from it only in one bit of the binary representations of the node indices see figure Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 50 Fig 4 1 Hypercube topology The hypercube topology template is the following it can be placed into a separate file e g hypercube ned Simple Node gates out out in in endsimple module Hypercube parameters dim nodetype submodules node nodetype 2 dim like Node gatesizes out dim in dim connections for i 0 2 dim 1 j 0 dim 1 do node i out j gt node i 2 j in j is bitwise XOR endfor endmodule When you create an actual hypercube you substitute the name of an existing module type e g Hypercube_PE for the nodetype parameter The module type implements the algorithm the user wants to simulate and it must have the same gates that the Node type has The topology template code can be used through importing the file import hypercube ned simple Hypercube_PE gates out out endsimple network hypercube Hypercube parameters dim 4 nodetype Hypercube_PE endnetwork If you put the nodet ype parameter to the ini file you can use the same simulation model to test e g several routing algorithms in a hypercube each algorithm implemented with a different simple module type you just have to supply different va
63. can get a more complete list In the commercial category OPNET is widely held to be the state of the art in network simulation OMNeT is targeted at roughly the same segment of network simulation as OPNET Seven issues are examined to get an overview about the network simulation tools Detail Level Does the simulation tool have the necessary power to express details in the model In other words can the user implement arbitrary new building blocks like in OMNeT or he is confined to the predefined blocks implemented by the supplier Some tools like COMNET III are not programmable by the user to this extent therefore they cannot be compared to OMNeT Specialized network simulation tools like NS for IP and CLASS for ATM also rather fall into this category Available Models What protocol models are readily available for the simulation tool On this point non commercial simulation tools cannot compete with some commercial ones especially OPNET which have a large selection of ready made protocol models OMNeT is no exception Defining Network Topology How does the simulation tool support defining the network topology Is it possible to create some form of hierarchy nesting or only flat topologies are supported Network simulation tools naturally share the property that a model network consists of nodes blocks entities modules etc connected by links channels connections etc Many commercial simulators have graphical
64. derived from cObject See the class library reference for more info about them 6 12 The parameter class cPar 6 12 1 Basic usage cPar is a Class that was designed to hold a value The value is numeric long or double in the first place but string pointer and other types are also supported cPar is used in OMNeT in the following places e as module parameters e as message parameters There are many ways to set acPar s value One is the set Value member functions cPar pp pp pp setDoubleValue 1 0 or by using overloaded operators cPar pp pp pp 1 0 For reading its value it is best to use overloaded type cast operators double dl double pp or simply double d2 pp Long integers pp 89363L or pp setLongValue 89363L Character string pp hi there or pp setStringValue hi there Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 122 The cPar object makes its own copy of the string so the original one does not need to be preserved Short strings less than 20 chars are handled more efficiently because they are stored in the object s memory space and are not dynamically allocated There are several other types cPar can store such as boolean void pointer cObject pointer function with constant args they will be mentioned in the next section For numeric and string types an input flag can be set
65. double max virtual double mean virtual double stddev virtual double variance Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 249 virtual void saveToFile FILE virtual void loadFromFile FILE Redefined cSt dDev functions virtual double random cWeightedStdDev s random number generator returns numbers of normal distribution with the current mean and standard deviation 15 12 4 class cDensityEstBase Common base class for density estimation classes Provides several pure virtual functions so it is an abstract class no instances can be created For the histogram classes you need to specify the number of cells and the range Range can either be set explicitly or you can choose automatic range determination Automatic range estimation works in the following way 1 The first few num_firstvals observations are stored 2 After having collected a given number of samples the actual histogram is set up The range min max of the initial values is expanded range_ext_factor times and the result will become the histogram s range rangemin rangemax Based on the range the cells are layed out Then the initial values that have been stored up to this point will be transferred into the new histogram structure and their store is deleted this is done by the transform function You may also explicitly specify t
66. function If you use handleMessage place the WATCH statement into initialize WATCH creates a dynamic cWatch object and we do not want to create a new object each time handleMessage is called 6 16 3 Snapshots The snapshot function outputs textual information about all or selected objects of the simulation including the objects created in module functions by the user into the snapshot file bool snapshot cObject obj amp Simulation const char label NULL The function can be called from module functions like this dump the whole network this dump this simple module and all its objects amp putAsideQueue dump queue contents amp Ssimulation msgQueue dump future events snapshot snapshot snapshot snapshot ee ee ee This will append snapshot information to the end of the snapshot file The snapshot file name has an extension of sna default is omnetpp sna Actual file name can be set in the config file The snapshot file output is detailed enough to be used for debugging the simulation by regularly calling snapshot one can trace how the values of variables objects changed over the Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 139 simulation The arguments label is a string that will appear in the output file obj is the object whose inside is of interest By default the whole simulation all modules etc will be
67. functions would add a virtual table pointer to objects Example usage opp_string a string b foo a bar a b const char s a printf string s n a Member functions opp_string opp_string const char s opp_string String creation destruction operator const char Returns pointer to the string const char operator const char s Deletes the old value and opp_st rdup s the new value to create the object s own copy 15 3 Declaration registration macros Header file ct ypes h sim directory Network NAME SETUPFUNCTION Network declaration macro it can be found in code generated by the NEDC compiler The use of this macro allows the creation of a network when only its name is available as a string Typically the name of the network to be simulated is read from the configuration file The macro expands to the definition of a cNetworkType object Define_Link NAME DELAY ERROR DATARATE Link type definition The macro expands to the definition of a cLinkType object the last three arguments are pointers to functions which dynamically create cPar objects an return their pointers Define_Function FUNCTION ARGCOUNT Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 214 Registers a mathematical function that takes 0 1 2 or 3 double arguments and returns a dou
68. generate NED files Text processing languages like perl and awk are great tools for that b write the network building code in C You can look at the output of nedc for some idea how to do it Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix B Parsec and OMNeT 198 13 APPENDIX B PARSEC AND OMNET 13 1 What is PARSEC PARSEC is a very successful simulation language with strong support for parallel simulation PARSEC bears some similarity to OMNeT in that it is also based on threads coroutines The language and the software has been developed at the Parallel Computing Laboratory of the University of California Los Angeles UCLA under the leadership of Prof Rajive Bagrodia PARSEC has been used in a number of simulation projects for example in simulation of mobile radio networks in a military environment It is best to quote the PARSEC User Manual Release 1 1 August 1998 PARSEC for PARallel Simulation Environment for Complex programs is a C based discrete event simulation language It adopts the process interaction approach to discrete event simulation An object also referred to as a physical process or a set of objects in the physical system is represented by a logical process a thread roughly equivalent to an OMNeT simple module Andras Interactions among physical processes events are modeled by timestamped message exchanges among the corresponding logical proce
69. have a regular structure three of them are described below Subgraph of a Full Graph This pattern takes a subset of the connections of a full graph A condition is used to carve out the necessary interconnection from the full graph for i 0 N 1 j 0 N 1 do node i out gt node j in if condition i j endfor The RandomGraph compound module presented earlier is an example of this pattern but the pattern can generate any graph where an appropriate condition i j can be formulated For example when generating a tree structure the condition would return whether node j is a child of node i or vica versa Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 49 Though this pattern is very general its usage can be prohibitive if the N number of nodes is high and the graph is sparse it has much fewer connections that N The following two patterns do not suffer from this drawback gt Connections of Each Node The pattern loops through all nodes and creates the necessary connections for each one It can be generalized like this for i 0 Nnodes j 0 Nconns i 1 do node i out j gt node rightNodeIndex i j in j endfor The Hypercube compound module to be presented later is a clear example of this approach BinaryTree can also be regarded as an example of this pattern where the inner j loop is unrolled The applicability of this pattern depends on
70. if you use an iterator class chOueueIterator you can examine each object in the queue The cQueuelterator constructor takes two arguments the first is the queue object and the second one specifies the initial position of the iterator 0 tail 1 head Otherwise it acts as any other OMNeT iterator class you can use the and operators to advance it the operator to get a pointer to the current item and the end member function to examine if youre at the end or the beginning of the queue An example for cQueuelterator iter queue 1 iter end itertt cMessage msg cMessage iter J hots Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 120 6 10 2 Expandable array cArray Basic usage cArray is a container class that holds objects derived from cObject cArray stores the pointers of the objects inserted instead of making copies cArray works as an array but if it gets full it grows automatically Internally cArray is implemented with an array of pointers if the array gets full it is reallocated cArray objects are used in OMNeT to store parameters attached to messages and internally for storing module parameters and gates Creating an array cArray array array Adding an object at the first free index cPar p new cPar par int index array add p Adding an object at a given index if the index is occup
71. include lt string h gt include lt time h gt include omnetpp h derive SmartPlayer from cSimpleModule class SmartPlayer public cSimpleModule Module_Class_Members SmartPlayer cSimpleModule 8192 this is a macro it expands to constructor definition etc 8192 is the size for the coroutine stack in bytes virtual void activity this redefined virtual function holds the algorithm register the simple module class to OMNeT Define_Module SmartPlayer define operations of SmartPlayer void SmartPlayer activity Technical University of Budapest Dept of Telecommunications OMNeT Manual An Example The NIM Game 30 int move initialization phase send module type to Game module create a message with the name SmartPlayer and send it to Game cMessage msg new cMessage SmartPlayer send msg out infinite loop to process moves simulation will be terminated by Game for 77 messages have several fields here we 1ll use the messag kind member to store the number of sticks cMessage msgin receive receive message from Game int num_sticks msgin gt kind extract message kind an int this hold the number of sticks still on the stack delete msgin dispose of the message move num_sticks 4 5 calculate move if move 0 we cannot take zero mov
72. is done via the member functions of ev Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 254 15 17 1 class cEnvir Construction destruction cEnvir cEnvir The constructor and the destructor in most cases do nothing User interface functions void inspect cObject object Creates an inspector window for the object void objectDeleted cObject object Notifies the environment that the object no longer exists The user interface should close all inspector windows for the object and remove it from object lists currently displayed cObject s destructor automatically calls this function void messageSent cMessage amp msg Notifies the environment that a message has been sent The user interface then can display the message in a text window animate the message s travel in the network s drawing activate conditional breakpoints and so on This function is automatically called right after cSimpleModule ssend placed the message into the future events queue Module function I O void printfmsg const char fmt Displays a message in dialog box This function should not be used too much by simple modules if ever void printf const char fmt n Simple modules can output text into their own window through this function The text is expected in printf format format string arguments void puts const char s Similar to cEn
73. it reaches its destination it is not scheduled individually for each link but rather every calculation is done once within the send call This implementation was chosen because of its run time efficiency In the actual implementation of queuing the messages at busy gates and modeling the transmission delay messages do not actually queue up in gates gates do not have internal queues Instead as the time when each gate will finish transmission is known at the time of sending the message the arrival time of the message can be calculated in advance Then the message will be stored in the event queue FES until the simulation time advances to its arrival time and it is retrieved by its destination module TBD add pseudocode Consequence The implementation has the following consequence If you change the delay or the bit error rate or the data rate of a link during simulation the modeling of messages sent just before the parameter change will not be accurate Namely if link parameters change while a message is under way in the model that message will not be affected by the parameter change although it should However all subsequent messages will be modelled correctly Similar for data rate if a data rate changes during the simulation the change will affect only the messages that are sent after the change If it is important to model gates and channels with changing properties you can go two ways e write sender module such tha
74. keyword italic identifier networkdescription definition definition include channeldefinition simpledefinition moduledefinition networkdefinition include INCLUDE fileName channeldefinition CHANNEL channeltype DELAY numericvalue ERROR numericvalue DATARATE numericvalue ENDCHANNEL KKK KKK Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix C NED Language Grammar Simpledefinition SIMPLE simplemoduletype machineblock paramblock gateblock ENDSIMPLE simplemoduletype moduledefinition MODULE compoundmoduletype machineblock paramblock gateblock submodblock connblock ENDSIMPLE compoundmoduletype moduletype simplemoduletype compoundmoduletype machineblock MACHINES machine paramblock PARAMETERS parameter parameter i parametername parametername CONST NUMERIC parametername STRING parametername BOOL parametername CHAR parametername ANYTYPE gateblock GATES IN gate OUT gate gate gatename submodblock SUBMODULES submodule submodule submodulename moduletype vector on_block substparamblock gatesizeblock submodulename parametername vector LIKE moduletype on_block substparamblo
75. lt lt n cMyClass amp cMyClass operator cMyClass amp myclass cObject operator myclass samples myclass samples See the virtual functions of cOb ject in the class library reference for more information 6 16 Tracing and debugging aids 6 16 1 Displaying information about module activity The global object called ev represents the user interface of the simulation program You can send data to ev using the C style I O operator lt lt ev lt lt started n ev lt lt about to send message lt lt i lt lt endl ev lt lt queue full discarding packet n The exact way messages are displayed to the user depends on the user interface In the command line user interface Cmdenv it is simply dumped to the standard output This output can also be disabled from the ini file so that it doesn t slow down simulation when it is not needed In windowing user interfaces Tkenv each simple module can have a separate text output window The above means that you should not use printf cout lt lt and the like because with Tkenv their output would appear in the xterm window behind the graphical window of the simulation application The user can also specify a phase string that is displayed at the top of the text output windows The phase string can indicate what the module is currently doing setPhase starting up for setPhase idle REIT se
76. network description actual module types can be parameters 2 2 2 Object mechanisms The use of smart container classes allows the user to build aggregate data structures For example one can add any number of objects to a message object as parameters Since the added objects can contain further objects complex data structures can be built Technical University of Budapest Dept of Telecommunications OMNeT Manual Overview 24 There is an efficient ownership mechanism built in The user can specify an owner for each object then the owner object will have the responsibility of destroying that object Most of the time the ownership mechanism works transparently ownership only needs to be explicitly managed when the user wants to do something non typical The forEach mechanism allows one to enumerate the objects inside a container object in a uniform way and do some operation on them This feature which makes it possible to handle many objects together The forEach feature is extensively used by the user interfaces with debugging capability and the snapshot mechanism see later 2 2 3 Derive new classes It most cases the functionality offered by the OMNeT classes is enough for the user But if it is needed one can derive new classes from the existing ones or create entirely new classes For flexibility several member functions are declared virtual When the user creates new classes certain rules need to be kept so that the obje
77. no way to know how many connections there1l be simultaneously The solution is to create a manager module which receives connection requests and creates a module for each connection The Dyna example simulation does something like this It is often convenient to use direct message sending with dynamically created modules Module factories TBD cModuleType moduleType findModuleType TCPConnectionHandler Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 111 Simple form Mainly for creating simple modules TBD cModuleType has createScheduleInit const char name cModule parentmod convenience function to get a module up and running in one step mod modtype gt createScheduleInit name this Does create buildInside callInitialize scheduleStart now Should work for both simple and compound modules Not applicable if the module e has parameters to be set e has gate vector sizes to be set e has gates to be connected before initialize Example TBD Expanded form If the previous simple form cannot be used There are 5 steps Step 1 find descriptor object Step 2 create module Step 3 set up parameters and gate sizes if needed Step 4 call function that builds out submodules and finalizes the module Step 5 call function that creates activation message s for the new simple module s Each step except for Step 3 can be done with one
78. of the Statistical Synchronization Method SSM Similarly to other parallel discrete event simulation methods the model to be simulated which is more or less a precise representation of a real system is divided into segments where the segments usually describe the behaviour of functional units of the real system The communication of the segments can be represented by sending and receiving various messages The simulators of the segments are executed by separate processors The communication of these segments is simulated with appropriate interfaces The messages generated in a given segment and to be processed in a different segment are not transmitted there but the output interfaces collect the statistical data of them If the input interfaces generate messages for the segments according to the statistical characteristics of the messages collected by the proper output interfaces the segments with their input and output interfaces can be simulated separately giving statistically correct results The events in one segment have not the same effect in other Technical University of Budapest Dept of Telecommunications OMNeT Manual Parallel Execution 179 segments as in the original model so the results collected during the SSM are not exact The precision depends on the segmentation on the accuracy of statistics collection and regeneration and on the frequency of the statistics exchange among the processors Segmentation The segm
79. omnetpp ini r e Machines nodel whale hit bme hu node4 whale hit bme hu node2 puppis hit bme hu node3 dolphin hit bme hu reee 10 1 5 Configuration of the slaves Slave processes can be configured in the Slaves section of the configuration file file omnetpp ini Slaves write slavelog slavelog file module messages errmsgs to console infomsgs to console modmsgs to console Screen input output of the slaves is re routed to the console However any file I O is done in the local file system of each host 10 1 6 Configuration and running The user must have PVM installed on the hosts he is going to run segments on To set up a simulation for distributed execution the user must 1 set the PVM_ROOT environment variable 2 link the simulation executable with sim_pvm instead of sim_std You can do it by setting PVM_SUPPORT to yes inamakemake generated makefile 3 set distributed true inthe General section of the configuration file Technical University of Budapest Dept of Telecommunications OMNeT Manual Parallel Execution 178 4 specify the logical hosts to physical machines mapping in the Machines section 5 copy the simulation executable and the configuration file to each host if they have physically different disks 6 start pvm with an appropriate pvmhosts file 7 start the simulation executable on the host which is supposed to be the console That proc
80. public cSimpleModule public void argmem The parent module code would be childmod gt argmem argument_memory_ptr sendDirect msg childmod 0 0 in Child module code would be msg receive argument_memory_ptr argmem op_pro_valid Given the module id int valid id gt 0 amp amp simulation exist id wl aa Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix A OPNET and OMNeT 197 Environment files Configuration files Default is onnetpp ini Multiple ini files and ini file inclusion are also supported Process Editor Your favourite text editor Or vi Network Editor Node Editor Any editor to write NED files GNED Not very sophisticated yet though Simulation Tool Use the Run 1 Run 2 etc sections in omnetpp ini do describe several runs with different parameters To create loops on different variables you can use a shell script that creates a short ini file with the variable parameters and include that file in omnnetpp ini probes Probe Editor From the ini file you can turn on off cOut Vector objects individually as well as assign result collection interval to them Analysis Tool Plove EMA Where you would normally use EMA OMNeT NED files with parameterized topology are often enough Otherwise you have two choices a write a program to
81. recursively insert starter messages for all submodules using activity do callnitialize on system module enter event loop described earlier if event loop terminated normally i e not with a runtime error do callFinish on system module clean up callnitialize call to user defined initialize function if module is compound for each submodule do callnitialize on submodule callFinish if module is compound for each submodule do callFinish on submodule call to user defined finish function initialize vs constructor Usually you should not put simulation related code into the simple module constructor For example modules often need to investigate their surroundings maybe the whole network at the beginning of the simulation and save the collected info into internal tables Code like that cannot be placed into the constructor since the network is still being set up when the constructor is called Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 80 finish vs destructor Keep in mind that finish is not always called so it isn t a good place for cleanup code which should run every time the module is deleted finish is only a good place for writing statistics result post processing and other stuff which are to run only on successful completion Cleanup code should go into the destructor But in fact you almost never need to write a de
82. seeds that greatly speeds up the tool For practical use the g option is the most important Suppose you have 4 simulation runs that need two independent random number generators each and you want to start their seeds at least 10 000 000 values apart The first seed value can be simply 1 You would type the following command C OMNETPP UTILS gt seedtool g 1 10000000 8 The program outputs 8 numbers that can be used as random number seeds 1768507984 33648008 1082809519 703931312 1856610745 784675296 426676692 110064264 You would specify these seed values in the ini file 8 4 Repeating or iterating simulation runs TBD Intro and multiple simulation runs in omnetpp ini vs controlling script Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 164 Variations over parameter values You don need to generate the whole omnetpp ini from program if you use include files You can have a fixed omnetpp ini which contains the line include parameters ini and then generate parameters ini by program for each run Here s the runall script of Joel Sherrill sFile System Simulator as an example bin bash This script runs multiple variations of the file system simulator all_cache_managers NoCache FIFOCache LRUCache PriorityLRUCache all _schedulers FIFOScheduler SSTFSch
83. simple modules using activity are booted by events starter messages inserted into the FES by the simulation kernel before the start of the simulation How does the coroutine know it has finished processing the event The answer when it requests another event The functions which request events from the simulation kernel are the receive family and wait so their implementations contain a transferTo main call somewhere Their pseudocode as implemented in OMNeT receiveNew other receive variations are similar transferTo main retrieve current event return the event remember events messages Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 72 wait create an event e and schedule it at current sim time wait interval while true transferTo main retrieve current event if current event is e break from loop else store current event for later use in the put aside queue delete event e return Thus the receive and wait calls are special points in the activity function because that s where e simulation time elapses in the module and e other modules get a chance to execute Starter messages Modules written with activity need starter messages to boot These starter messages are inserted into the FES automatically by OMNeT at the beginning of the simulation even before the initialize functions are called
84. stuff in the NED file simple modules channels networks etc will survive the operation GNED puts all graphics related data into display strings Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 59 GNED works by parsing your NED file into an internal data structure and regenerating the NED text when you save the file One consequence of this is that indentation will be canonized hopefully you consider this fact as a plus and not as a minus Comments in the original NED are preserved the parser associates them with the NED elements they belong to so comments wont be messed up even if you edit the graphical representation to death by removing adding submodules gates parameters connections etc GNED is now a fully two way visual tool While editing the graphics you can always switch to NED source view edit in there and switch back to graphics Your changes in the NED source will be immediately backparsed to graphics in fact the graphics will be totally reconstructed from the NED source and the display strings in it GNED is still under development There are some missing functions and bugs but overall it should be fairly reliable See the TODO file in the GNED source directory for problems and missing features Comment parsing It is useful to know how exactly GNED identifies the comments in the NED file The following maybe a bit long NED code should explain it
85. the directory where you want to install OMNeT and extract the archive using the command tar zxvf omnetpp tgz A subdirectory called something like omnetpp 2 0 will be created which will contain the simulator files In this text we ll assume this directory isc omnetpp 2 0 Make sure the MSVC executables nmake cl link are in the path It is also recommended that you put the c omnetpp 2 0 bin directory into the path Check the configuser vc file to make sure it contains the proper settings You ll probably need to correct the paths in the following lines OMNETPP_ROOT c omnetpp 2 0 TK_INCL_DIR c tcl include TK_LIB_DIR c tcl lib and maybe also the file names in the definition a few lines further below WISH S TK_LIB_DIR bin wish82 exe TK_LIBS tcl82 1lib tk82 1lib libpath TK_LIB_DIR Then type nmake f Makefile vc This should build the executables and the libraries and copy them to the bin and 1ib subdirectories within the top level OMNeT directory 7 4 3 Building the samples from the MSVC IDE Unfortunately MSVC doesnt like the cc extension so first you have to rename the cc files to cpp You can do that with samples cc2cpp bat Technical University of Budapest Dept of Telecommunications OMNeT Manual Building Simulation Programs 153 To link executables with Tkenv the MSVC project files need the location of the Tc
86. the following functions msg gt setKind kind msg gt setLength length msg gt setPriority priority msg gt setBitError err msg gt setTimestamp msg gt setTimestamp simtime With these functions the user can set the message kind the message length the priority the error flag and the time stamp The set TimeStamp function without any argument sets the time stamp to the current simulation time The values can be obtained by the following functions int k msg gt kind int p msg gt priority int 1 msg gt length bool b msg gt hasBitError simtime_t t msg gt timestamp Duplicating messages It is often needed to duplicate a message for example send one and keep a copy This can be done in the standard ways as for any other OMNeT object cMessage copyl cMessage msg gt dup cMessage copy2 new cMessage msg The two are equivalent The resulting message is an exact copy of the original including message parameters cPar or other object types and encapsulated messages 6 3 2 Attaching parameters and objects to a message Adding setting and reading parameters You can add any number of parameters to a cMessage object Parameters are objects of cPar type You add a new parameter to the message with the addPar member function msg gt addPar dest_addr You can get back the reference to the parameter object with the par
87. the ini file see later Stopping on errors If your simulation detects an error condition and wants to stop the simulation you can do it with the error member function of cModule It is used like printf if windowSize lt 1l error Invalid window size d must be gt 1 windowSize Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 106 Do not include a newline n or punctuation period or exclamation mark in the printed out text it will be added by OMNeT 6 5 Accessing module parameters and gates 6 5 1 Module parameters Module parameters can be accessed with the par member function of cModule cPar amp delay_par par delay The cPar class is a general value storing object It supports type casts to numeric types so parameter values can be read like this int num_tasks par num_tasks double proc_delay par proc_delay If the parameter is a random variable or its value can change during execution it is best to store a reference to it and re read the value each time it is needed cPar amp wait_time par wait_time for hehe ss wait simtime_t wait_time If the wait_t ime parameter was given a random value e g exponential 1 0 in the NED source or the ini file the above code results in a different delay each time Parameter values can also be changed from the program during execution
88. the sequence If you have several runs each run is started with a fresh set of seeds that are 1 000 000 values apart from the seeds used for previous runs Since the generation of new seed values is costly OMNeT has a table of precalculated seeds 256 values if they are all used up OMNeT starts from the beginning of the table again Automatic seed selection may not be appropriate for you for several reasons First you may need more than 256 seeds values or if you use variance reduction techniques you may want to use the same seeds for several simulation runs In this case there is a standalone program to generate appropriate seed values seedtool will be discussed in the next section and you can specify the seeds explicitly in the ini file The following ini file explicitly initializes two of the random number generators and uses different seed values for each run Run 1 gen0 seed 1768507984 genl seed 33648008 Run 2 gen0 seed 1082809519 genl seed 703931312 If you want the same seed values for all runs you will write something like this General gen0O seed 1768507984 genl seed 33648008 All other random number generators 2 3 will have their seeds automatically assigned As a third way you can also set the seed values from the code of a simple module using genk_randseed but I see no reason why you would want to do so 8 2 8 List of all ini file options The exact meaning of the different
89. the total stack size sum of all coroutine stacks in kilobytes You need to increase this value if you get the Cannot allocate coroutine stack error load libs Name of shared libraries so files to load after startup You can use it to load simple module code etc Example load libs x25 x25 so lapb lapb so netif check freq used with parallel execution gen0 seed 3567 Seeds for the given random number generator They genl seed 4535 should not be zero Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 162 Cmdenv runs to execute 1 3 4 6 Specifies which simulation runs should be executed module messages yes no Globally enables disables ev style messages in simple modules e g ev lt lt sending n Verbose simulation yes no Enable disable banners for each event Event 1234 T stuff display update 100ms If there would be no display from the simulation execution both the above options are disabled OMNeT can print out regular messages of the progress The interval is understood in simulation time Tkenv default run 1 Specifies which run Tkenv should set up automatically after startup If there s no default run entry or the value is 0 Tkenv will ask which run to set up use mainwindow yes print banners yes breakpoints enabled yes Specifies whether the simulation should be
90. to code the function in C The C function must take 0 1 2 or 3 arguments of type double and return a double The function must be registered in one of the C files with the Define Function macro An example function the following code must appear in one of the C sources double average double a double b Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 56 return atb 2 Define_Function average 2 The number 2 means that the average function has 2 arguments After this the average function can be used in NED files module Compound parameter a b submodules proc Processor parameters av average a b endmodule An important application of this concept is to extend OMNeT with new distributions 4 9 8 Display strings Display strings specify the arrangement and appearance of modules in graphical user interfaces currently only Tkenv they control how the objects compound modules their submodules and connections are displayed Display strings occur in NED description sdisplay phrases The display string format is a semicolon separated list of tags Each tag consists of a key usually one letter an equal sign and a comma separated list of parameters like p 100 100 b 60 10 rect o blue black 2 Parameters may be omitted also at the end and also inside the parameter list like p 100 100 b rect o0 2 Module submodule param
91. to determine in how many hops the target node can be reached from our node through a particular output gate To calculate this you calculate the shortest paths to the target from the neighbor node but you must disable the current node to prevent the shortest paths from going through it sTopoNode thisnode topo nodeFor this thisnode gt disable topo unweightedSingleShortestPathsTo targetnode thisnode gt enable Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 116 for int j 0 j lt thisnode gt outLinks J sTopoLinkOut link thisnode gt out i ev lt lt Through gate lt lt link gt localGate gt fullName lt lt lt lt 1 link gt remoteNode gt distanceToTarget lt lt hops lt lt endl In the future other shortest path algorithms will also be implemented unweightedMultiShortestPathsTo sTopoNode target weightedSingleShortestPathsTo sTopoNode target weightedMultiShortestPathsTo sTopoNode target 6 9 Generating random numbers Random number generation is considered to be an important issue The random number generator used in OMNeT is a linear congruential generator LCG with a cycle length of 2 12 The startup code of OMNeT contains code that checks if the random number generator works OK so you do not have to worry about this if you port the simulator to a new architecture or use a different c
92. treated as mere pointer no memory management Duplication copies the pointer and deletion does nothing NULL NULL gt 0 size Plain memory management Duplication is done with new char size memcpy and deletion is done via delete NULL or user s indifferent Sophisticated memory management user s delete dupfunc Duplication is done by calling the user func supplied duplication function which should do the allocation and the appropriate copying Deletion is done by calling the user supplied delete function or the delete operator if it is not supplied Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 232 static int cmpbyvalue cObject one cObject other Compares two cPars by their value if they are numeric This function can be used to sort cPar objects in a priority queue Handling redirection cPar amp setRedirection cPar par TBD bool isRedirected TBD cPar redirection Returns NULL if the cPar s value is not redirected to another cPar otherwise it returns the pointer of that cPar This function is the only way to determine if an object is redirected or not type returns the type of the other cPar D L etc void cancelRedirection Break the redirection The new type will be long L Assignment and conversion operators This set of overloaded operators provides assignment
93. uaetuerenaruats 121 6 12 2 Random number generation through cPaf seeeseeeseesssesssereesseessresseesseesseee 122 6 12 3 Storing object and non object pointers in CPar ssssssesssesssesessseessresseesseseseee 122 6 12 4 Reverse Polish expressions seesssssesssessesssessseessstessttsseesseresseeessresseesseesseee 123 6 125 Using redirection aesae e a N a A a daa E duseseioneenate 124 6 12 6 Type chara ES a E E A E EASA 125 6 127 SUMMA ea na a a 125 6 13 Statistics and distribution estimation ssssssessseeessseesseessersseseseetssetesseessersseresseeessresseesse 127 6 13 1 cStatistic and descendants 452 0 cersaiticcptucgostes eens tragdanasacs tesssaeuaneuted ena aiese Gaeiaens 127 615 2 Distribution estimatio ies iret estee isre aa aaor tass ledges aE i Ss 128 6 13 3 The k split alg rithm nescion ens naa a aE S a EE 131 6 13 4 Transient detection and result accuracy essssesssseessessseresereesseessressersseesseee 133 6 14 Recording simulation results snseeseeesseesseesseeesseetsseesseessessereeseeessresseesseeesseesseesseesse 134 6 141 Output vectors cOutV COLOR cnni i e i acne vewee 134 6 14 2 Output Scalas nn onn a a a een E A r 135 6 15 Deriving NEw ClaSSES i sisiascisssdsavssieizaesdesecesasaniaseasvedaacnedanades detedacs E A E EENES Ea aas 135 6 16 Tracing and debugging AN Se 5 a5 sacar s cy acs nate ects es uae santas oy tastine sdaadeieoastace ees ecaanmoases 137 6 16 1 Displaying informa
94. verbose simulation no line The heapcheck defines were turned on all screen output on 2 min We turned off heapcheck we commented out the no heapcheck defines in heap cc and turned back on the screen output We used the same omnetpp ini setting as with first case screen output redir 15 5 sec Same as previous configuration except that we run the ected to file program with no heapcheck fddi gt output txt screen output redir 13 sec Same as previous configuration except that we run the ected to NUL program with no heapcheck fddi gt NUL screen output turned 7 5 sec We did not only redirect but also disabled screen off from ini file output Setting in omnetpp ini no heapcheck verbose simulation no module messages no no screen output 4 5 sec We ifdef ed out all printouts from the simple generation module sources and also turned off any messages from no heapcheck omnetpp ini The moral is that heap checks and screen output greatly influences speed so once you do not need them debugging is over throw them out You also gain a lot by putting ifdef lines around your debugging code And of course program with care Technical University of Budapest Dept of Telecommunications OMNeT Manual Analyzing Simulation Results 171 9 ANALYZING SIMULATION RESULTS 9 1 Plotting output vectors with Plove 9 1 1 Plove features Typically youll get output vector files as a result of a simulation
95. were owned by the queue will be deleted int length Returns the number of objects contained in the queue bool empty Returns t rue if the queue is empty void insert cObject obj Inserts the given object into the queue maintaining the sorting order void insertBefore cObject where cObject obj void insertAfter cObject where cObject obj Inserts exactly before and after the given object cObject tail Returns pointer to the last object in the queue or NULL if the queue is empty cObject pop Unlinks and returns the last object in the queue cObject remove cObject obj Unlinks and returns the object given 15 9 3 class cQueuelterator To examine each element in the queue the cOueueIterator iterator class can be used Once a cQueuelterator object is created for the queue the cQueue object the and operators can be used to step from one element of the list to the next previous one cOueueIterator is nota descendant of cObject or cIterator cQueuelterator cQueue amp q int a 1 Constructor cIterator will walk on the queue passed as argument The current object will be the first if a 1 or the last a 0 object in the queue void init cQueue amp q int a 1 Reinitializes the iterator object cObject operator Returns the current object bool end Returns t rue if we have reached either end of the queue cObject amp operator int cO
96. will be the first if a 1 or the last a 0 item in the list void init cLinkedList amp q int a 1 Reinitializes the iterator object void operator Returns the current item bool end Returns t rue if we have reached the end of the list void operator int void operator int Steps to the next previous object in the list 15 10 3 class cBag cBag is a container class which is designed to hold non cObject items without constructors destructors ints doubles small structs etc cBag stores full bit by bit copies of the items inserted and not the pointers cBag works as an array but if it gets full it grows automatically by a specified delta Construction destruction copying cBag cBag amp bag Copy constructor Bag const char name NULL int esiz l int siz 0 int delt 5 cHead h amp locals Constructor It takes the object name the size of an item in bytes the initial vector size delta and the pointer to the list it should join as argument Q virtual cBag Destructor The items in the object will be deleted virtual cObject dup cBag amp operator cBag amp bag Duplication and assignment work all right with cBag Redefined virtual functions virtual const char className j Returns a pointer to the class name string cBag virtual void info char buf virtual const char inspectorFactoryName Redefined
97. 0 15 2 Constants types simple functions eesesseeessesesseserssressessrestesserseesrersteseesrreseesseseresreses 210 15 2 T Constants types e a tan eat A eats 210 15 2 2 Function typedefs senene oea ee E eA a 210 15 2 3 Simulation time conversion functions ssssesssssesssesessseesstessresseesseeesseeessees 211 15 2 4 Value added string TUT CEONS 53 sii ing se isl so vas deed ah esta oes andes eos Geo 211 15 2 5 Basic random number SENET AON rsinohg asthe walein carte acied Gaasetiendamagos 211 15 2 6 Random numbers from different distributions eeeeeeeeeeeeeseeeeeteeeeneeees 212 W527 NSC NAME GUS ea A au eteeduae dat tem A tn tae Aenea 212 MD is SINE CLASS ec aise n Sadek ut sae weiss oi ce ds a aca e ee oe oua E aa aes 213 15 3 Declaration repistration MACOS 4 1iciecccsisssessacsateasacsaseseaceasvacecde todententveraccesndeeceasncosaterons 213 15 4 Common properties for all classes CODJjeCt es ceeececeeseecesececeeceeceeeeeceeeeeeseeeeneeeenaeeees 214 TS Al Class COB SCL Gece E es ica sca suse tags E E E une deaeoa ease ee heen aeases 215 15 5 Module classes cModule cCompoundModule cSimpleModule eeeeeeeeteeees 218 15 5 1 elassceMod le nsista e ias 218 15 5 2 class cCompoundModule sjss sseejaceiasnsucge tds d cena seas sangians evsendenie led naaiese ynesaess 221 1329 5 class CSP le NO del Sai e Geek Pease sabes Sarat cea iee eas e gei e a a 221 15 6 Message classes cMessage and CPacket x
98. 1 Statistics collection and run time analysis OPNET OMNeT writing observations to output file writing observations to output files probes to select statistics to be collected roughly equivalent to OPNET s solution only off line analysis analysis of output files is supported run time processing basic measures mean etc distribution estimation with histograms quantiles P algorithm support for detecting the end of the transient period and sufficient result accuracy Parallel execution OPNET OMNeT not supported supported over PVM arbitrary synchronization can be used Openness OPNET OMNeT Input file formats binary model files text files textual parameter files Output file formats binary files Availability of source not available only the available source of the shipped models is available Embedding simulations not supported and also not supported into other software possible the main product function cannot be supplied by the user etc Embedding application becomes a new user interface based on Envir 1 or embedding application replaces Envir 2 Can be read and analyzed by EMA programs Can be exported to text files from the main OPNET program 12 2 Quick reference for OPNET users This section is intended to help OPNET users learn OMNeT faster Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix A O
99. 1 6 Configuration and running ss ssessessseseseeeesseessetsserssereseeesseessesseesseeesseeessees 177 10 4 PEGG re are problem Sariei a S S 178 Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 10 102 Statistical SyNehroniZallOn sy seinni teens acansavbiaes E E aaeley asia 178 10 2 1 The description of the Statistical Synchronization Method SSM 178 10 2 Using SSM N OMNEeT FF oer cj dvestaggncaneavoeca sa datededes aia sehen ONG a ear h 179 11 THE DESIGN OF OMNET ccccssssccccccccccsssscccsccccccccccececcscsccccccccccececssccscececoees 180 11 1 Structure of an OMNeT executable ccccccecccecccccccccssssesesccccsseesssesesccesseseenenenss 180 Td Porabe dang OMNeET Fuen gal ate ee case te das ete aa es cat etna daloed 180 11 3 The simulation Kernel siiccs33ac ics sescchsteadocess aa ta ccdats aetiegstend diese settee belies Sebel tea Naa 181 11 3 1 The central object cSimulation simulation 0 0 0 cece eeeeeeneeceeeeeeeeeeeeeeneees 181 l3 2 Mody E Classes rcosics so3s05chkt0 hs siads oodhdate a a es 181 11 3 3 Global registration lists iccisscavsisisageadiagsavcavaleseaacessnszevasdecsdeadsenacessadenuenseessaneries 181 1173 4 The Coroutine packages ninien iaer i ae iai oni 183 11 3 5 Object ownership contains relationships esssssesssessseseesseesseessersserssseeessees 183 E E A EAE E T E E A E EAT 184 Z ANEO AETAT O E E E 184 11 4
100. 15 9 1 class cArray cArray is a container object that holds objects derived from cOb ject cArray stores the pointers of the objects inserted instead of making copies cArray works as an array but if it gets full it grows automatically by a specified delta Ownership of contained objects responsibility of deletion can be specified per object basis see cObject takeOwnership Default is that cArray takes the ownership of each object inserted that is takeOwnership true Construction destruction copying cArray cArray list Copy constructor Contained objects that are owned by cArray that is whose owner is the cArray will be duplicated so that the new cArray will have its own copy of them cArray const char name NULL int siz 0 int dt 10 cHead h amp locals Constructor It takes the object name the initial vector size delta and the pointer to the list it should join as argument virtual cArray Destructor The contained objects that were owned by the container will be deleted virtual cObject dup cArray operator cArray list Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 235 Duplication and assignment work all right with cArray Contained objects that are owned by cArray will be duplicated so that the new cArray will have its own copy of them Redefined virtual functions virtual const char className
101. 2 The CEnvir interface ier a a a a ESER NES 184 11 4 3 Implementation of the user interface simulation applications eee 184 11 5 Writing inspectors for TRETIV sso jeveskisasengsess aiesdagencgaads Uaeasasaasaxgaass cesedsaderes tha danke aaedevsanesake 185 12 APPENDIX A OPNET AND OMNET cccssssscsccccccccecsssssccscccccccccceccscccscsceosees 186 12 1 Comparison of OPNET and OMNeT 4 2 0 cee ccesscecsscceesscceessccessseceesscesensccsssecesenseeees 186 12 2 Quick reference for OPN ET Users nkiuteiieeei nai i aaaea 191 13 APPENDIX B PARSEC AND OMNET ccccsssssssscccccccceccscsssssccccccccececcccscccscecsees 198 B What e PARSE O r aa n r a R i 198 13 2 What is inside the Parsec package vss evaicisascatsassavsava bovsaccvssaceevasdevedeadeenaccedadenessenczanerens 198 13 3 Parsec vs the OMNeT simulation kernel c cece cesseseecscccccecesseeecescessssesneeeess 199 13 4 F at r summai snenge a e E e eeii os EST Eeer 203 13 5 Correspondence between Parsec and OMNeT sseessesseseeseeseresesreesresrsrresreseeseresresss 205 14 APPENDIX C NED LANGUAGE GRAMMAR sssesesesesosososesosososesosososesosososososesososo 206 15 APPENDIX D CLASS LIBRARY REFERENCE sesesseseseeseeceseseososeoeosessoseseosesessoseseesese 210 Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 11 15 Whatis document ed Nere o cctteseveseant seve ces dios taaductesis a a E 21
102. 5 window_size 2 connections for i 0 num_of_ports 1 do routing output i gt port_if i from_higher_layer routing input i lt port_if i to_higher_layer port_if i to_port gt output_ports i port_if i from_port lt input_ports i endfor routing output num_of_ports gt local_user input routing input num_of_ports lt local_user output endmodule Example 2 Chain For example one can create a chain of modules like this module Serial parameters count const submodules node Node count gatesizes in 2 out 2 gatesizes if index 0 index count l1 in L ouk hs connections for i 0 count 2 do node i out i 0 1 0 gt node itl in 0 node i in i 0 1 0 lt node itl out 0 endfor endmodule Example 3 Binary Tree Building a binary tree is a good example of using conditional connections simple BinaryTreeNode gates in from_up from_downleft from_downright out upward downleft downright endsimple module BinaryTree parameters height const submodules node BinaryTreeNode 2 height 1 Liars Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 47 connections for i 0 2 height 2 j 0 2 height 2 do node i upward gt node j from_downleft if leftchild i j node i from_up lt node j downleft if leftchild i j node i upward gt
103. Constructor It accepts the object name virtual cLinkedList Destructor calls clear virtual cObject dup cLinkedList amp operator cLinkedList amp llist Duplication and assignment work all right with cLinkedList Contained items are treated as configured with configPointer By default only pointers are copied Redefined virtual functions virtual const char className Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 240 Returns pointer to the class name string cLinkedList virtual void info char buf virtual const char inspectorFactoryName Redefined Member functions void configPointer VoidDelFunc delfunc Configures memory management for contained items Similar to cPar s configPointer 0 1 function delete func dupl func itemsize NULL NULL 0 NULL gt 0 size or user s indifferent user s delete dupfunc void clear As a result the container will be empty configPointer int length VoidDupFunc dupfunc behaviour Pointer is treated as mere pointer no memory management Duplication copies the pointer and deletion does nothing Plain memory management Duplication is done with new char size memcpy and deletion is done via delete Sophisticated memory management Duplication is done by calling the user supplied duplication function
104. Cs iol AER u U WAR AAR PAAR A P A P 888 AAA n WEPAAPAPAPRPPPP P i pA D A TECHNICAL UNIVERSITY OF BUDAPEST Faculty of Electrical Engineering and Informatics Department of Telecommunications OMNeT Discrete Event Simulation System Version 2 0b4 User Manual by Andr s Varga Last updated May 23 2000 OMNeT Manual Introduction 2 TOPICS 1 INTRODUCTION vacesusecs oncoesseasesusescecsavesnnesatecnsonsssdeconsssvesseveassoveds tuceaeveavenstseassadoueseacanss 14 De COVERY TE W soceakccesssencs so nsssecssiuees coatspeuessvedsssnasieseussabedassues ceaccesbed ensue iousoausnea nendsseacesstenas 19 3 AN EXAMPLE THE NIM GAME ssscscsacocsscsetsadsisccasssesedsonatocsocnoessies esscvechenecdoseassoseseos 27 4 THE NED LANGUAGBisccsisacsssetsssssicsontcensdssdvoseoseceenpssnsoosteosdtnn seen tscegusacoonsdessncoguoecuasens 35 5 SIMPLE MODULES iscisssesccticvspetbusseosestossassounsocusdousesnihobssussisessseussosseonbascssatissbeatooseoesets 61 6 THE SEMULATION LIBRARY sicctssesssetsscccesssssonstsaoncctonessosatcacebesvaesqeaseveenscoenstgousbeossdeees 92 7 BUILDING SIMULATION PROGRAMS e seessessessessesseesoesessoesoesoeecossessossossossoesoeesessesese 147 S RUNNING THE SIMULATION ccicoscousseosscducdesdsncsssussbeasaveseensnsustes toscanseasseedcossosssostosehge 155 9 ANALYZING SIMULATION RESULTS seiscscscesssossssssacsoedscssisedadesssssssstoc vedsedssdensesssecss 171
105. Data written to cOutVector objects from simple modules go to output vector files Normally you use Plove to look into output vector files and plot vectors in it Plove is a handy tool for plotting OMNeT output vectors It uses Gnuplot to do the actual work You can specify the drawing style lines dots etc for each vector as well as set the most frequent drawing options like axis bounds scaling titles and labels etc You can save the gnuplot graphs to files postscript latex pbm etc with a click Plove can also generate standalone shell scripts that plot output vectors in much the same way Plove does itself These scripts can be used for batch processing or to debug filters see later Plove does not take away any of gnuplot s flexibility you can embed your own gnuplot commands to customize the output Filtering the results before plotting is possible Filters can do averaging truncation of extreme values smoothing they can do density estimation by calculating histograms etc Some filters are built in and you can easily create new filters or modify the existing ones Filters can be incorporated in one of three ways as awk expressions as awk programs and as external filter programs Filters can be parameterized Multiple filters for the same vector is not currently supported also you cannot currently feed several vectors into a single filter Plove does not create temporary files so you don t need to worry about disk space if t
106. E hessssssisdoosoosooisioslsoveseostoora seice seere stias osiot seso os ae siseosas ees a a NED KODE A A RE R A te edie E FE ET 4 1 1 Components of a NED description ss seeessseesseesseessesesseeesseessresseessee 4 1 2 Reserved WOrdS aaea A a E O iR 4 1 3 Case sensitivi orea a a a O R R AD The imp rt statement innisin aen a n a E a 43 Channel definitions inene Ae Ree One EE Se eR NE Re E E 4 4 Simple module definitions sssessensseesseeeseeeesseessresseeeseeesseeesstesseesseesseeessseesseeso 4 4 1 Simple module parameters ceecceessecesneecesececeeececeeeeeceeeeeceteeeeeaeees 4 4 2 Simple module gates vse sesveisaveaceneisaieriuusaveisalajeadsnssnczevasdevedeadoonaacendenes 4 5 Compound module detmiliOns lt cicsieascnceanacesatesseaasareseasisedeanienetaaiencnetaeennias 4 5 1 Compound module parameters sssseesseeeesseesseesseesseeesretesseessresseessee 4 5 2 Compound module Cates scx scch castes as teasasceneaesdoesdosd oecatedadecentaneemeeeaieen 45 8 SUDIMGUUICS 4 6 cee oes eMedia ce eel dee oe AD AC ONMECHONS wsistes oi csatsseccseceien EEEE ETE EAE EE 4 6 Parameterized compound MOCUIES sc ccssssgrvocsansvacsactossaacaysencsdseshedessensengecdgadans BG bP ANS aenn e tema uaasensy ass E R 4 6 2 Using const with parameterized topolOgies eeeeeeceeceereeeeeeeeeeteees 4 6 3 Design patterns for compound Modulles ceeeececeeeeeeeeeeeeeeeeeeeees 4 6 4 Topology templates ssc
107. EC has a very efficient support for parallel execution Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix B Parsec and OMNeT 199 13 3 Parsec vs the OMNeT simulation kernel In the further sections I will give a brief overview of Parsec with special attention to the strong and weak sides compared to OMNeT NOTE I am doing my best however THIS IS NOT A REPLACEMENT FOR THE PARSEC MANUAL It is possible that there are errors in this chapter If you want to have a closer look at Parsec you must download and read the official Parsec manual from UCLA As it was mentioned earlier we can only compare Parsec against the OMNeT simulation kernel that is the functionality around send receive wait and the coroutines the activity function because Parsec simply doesn t offer much more than that The first look The PARSEC is a programming language based on C not C PARSEC programs in addition to normal C code contain special syntactic constructs so they do not compile as C You have to invoke the PARSEC compiler pcc which parses the whole program and translates it into true C code The main advantage of this solution is that the PARSEC language is clean and really elegant Let us see a bit of PARSEC code include lt stdio h gt message job int id int count z message add_to_your_sorc ename id entity driver int argc char argv
108. FSM support is described in the next section 5 3 3 initialize and finish Purpose initialize to provide place for any user setup code finish to provide place where the user can record statistics after the simulation has completed When and how they are called The initialize functions of the modules are invoked before the first event is processed but after the initial events starter messages have been placed into the FES by the simulation kernel Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 79 Both simple and compound modules have initialize functions A compound module has its initialize function called before all its submodules have The finish functions are called when the event loop has terminated and only if it terminated normally i e not with a runtime error The calling order is the reverse as with initialize first submodules then the containing compound module The bottom line is that in the moment there s no official possibility to redefine initialize and finish for compound modules the unofficial way is to write into the nedc generated C code Future versions of OMNeT will support adding these functions to compound modules This is summarized in the following pseudocode although you won find this code as is in the simulation kernel sources perform simulation run build network i e the system module and its submodules
109. FactoryName used by Tkenv to create inspector windows for objects of this type One should also use the Register_Class macro to register the new class It is used by the createOne function and the PVM extension of OMNeT Let us see a simple example The header file File cmyclass h include cobject h class cMyClass public cObject public int samples cMyClass cMyClass amp myclass cMyClass const char name NULL int k 0 virtual cMyClass virtual const char className return cMyClass virtual cObject dup return new cMyClass this virtual void info char buf virtual void writeContents ostream amp os cMyClass amp operator cMyClass amp myclass The corresponding cc file File cmyclass cc include lt stdio h gt include lt string h gt include lt iostream h gt include cmyclass h Register_Class cMyClass cMyClass cMyClass cMyClass amp myclass cObject setName myclass name operator myclass cMyClass cMyClass const char name int k cObject name samples k Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 137 void cMyClass info char buf cObject info buf sprintf buf strlen buf samples d samples void cMyClass writeContents ostream amp os os lt lt samples lt lt samples
110. OMNeT Portable User Interface for the OMNeT Simulation System Diploma Thesis Technical University of Budapest 1994 in Hungarian OMNS Lencse Gabor Graphical Network Editor for OMNeT Diploma Thesis Technical University of Budapest 1994 in Hungarian OMN6 Varga Andras OMNeT Portable Simulation Environment in C TDK work Technical University of Budapest 1992 in Hungarian Other simulation software See web siteTBD C language Too many books to list Cyg Win32 CYGWIN http sourceware cygnus com cygwin top html DJGPP DJGPP1 Official DJGPP Home Page http www delorie com djgpp PVM PVM 1 The Official PVM Home Page http www epm ornl gov pvm pvm_home html1 PVM2 http www sp2 uni c dk PVM PvmIntro html PVM3 http www cse ogi edu DISC projects mist related work pvm html Turbo Vision TV 1 Borland C 3 1 Manuals Borland International 1992 TV2 The TVPlus Archieve http wvnvm wvnet edu u6ed4 tvhome htm TV3 Sierwald Joern 32 bit Portable Turbo Vision http wvnvm wvnet edu ubed4 tvptsier htm TCL TK TCLTK1 Welch Brent Practical Programming in Tcl and Tk Prentice Hall 1995 Technical University of Budapest Dept of Telecommunications OMNeT Manual References 259 TCLTK2 TCLTK3 Gnuplot GPLOT1 GPLOT2 PMTV1 Xmer XMGR1 HyperTcl http web cs ualberta ca wade HyperTcl TCL WWW Info http www s
111. OMNeT Manual Simple Modules 74 5 3 2 handleMessage Function called for each event The idea is that at each event we simply call a user defined function instead of switching to a coroutine that has activity running in it The user defined function is the handleMessage cMessage msg virtual member function of cSimpleModule the user has to redefine the function to make it do useful work Calls to handleMessage occur in the main stack of the program no coroutine stack is needed and no context switch is done The handleMessage function will be called for every message that arrives at the module The function should process the message and return immediately after that The simulation time is potentially different in each call No simulation time elapses within a call to handleMessage The pseudocode of the event loop which is able to handle both activity and handleMessage simple modules while FES not empty and simulation not yet complete retrieve first event from FES t timestamp of this event m module containing this event if m works with handleMessage m gt handleMessage event else m works with activity transferTo m Modules with handleMessage are NOT started automatically the simulation kernel creates starter messages only for modules with activity This means that you have to schedule self messages from the initialize function if you want the handleMessage simple module t
112. PNET and OMNeT 192 OPNET OMNeT network subnetwork node module process interrupts invocations states Compound modules An OMNeT simple module corresponds to an OPNET module with its process When using handleMessage interrupt event invocation call to handleMessage state FSM state or the value of the state vars stored in the class When using modules with activity this means a little different way of thinking from OPNET s In OMNeT 4 you write a simple module as you would write an operating system process or a thread thus there s no need to distinguish states or speak about invocations Within the simulation kernel an invocation corresponds to a transferTo module call An OMNeT module accepts messages and simulation time advances within receive calls wait is just a scheduleAt followed by a receive An OPNET interrupt is the event being processed In this sense OMNeT messages returned by receive correspond to OPNET interrupts endsim interrupt op_ima_obj_attr_get op_ima_sim_attr_get The finish virtual member functions of the simple modules are called at the end of the simulation run You can redefine finish to write statistics etc foo par foo foo module gt par foo There are no simulation attributes You can use the parameters of the top level module instead op_sim_end
113. Reference 212 Range E22 Period length 2 12 Method x x 7 mod 27 1 To check if it works correctly starting with x 0 1 then x 10000 1 043 618 065 must hold Required hardware exactly 32 bit integer arithmetics Source Raj Jain The Art of Computer Systems Performance Analysis John Wiley amp Sons 1991 pages 441 444 455 You can use the generator through the following functions void opp_randomize Sets the random number generator s seed to a random value long randseed long seed 0 Sets the random number generator s seed to the given value Zero cannot be used as seed long intrand Returns an integer random number in the range 0 INTRAND_MAX 1 long intrand long r Returns an integer random number in the range 0 r 1 works if r lt lt INTRAND_MAX double dblrand Returns a random number in the range 0 0 1 0 15 2 6 Random numbers from different distributions double uniform double a double b Uniform distribution in the range a b double intuniform double a double b An integer with uniform distribution in the range a b Note that the function can return also a or b double exponential double mean Exponential distribution with parameter mean double normal double mean double variance Normal distribution with mean and variance given double truncnormal double mean double variance Normal distribution truncated to nonnegative values
114. Returns pointer to the class name string cArray virtual void info char buf virtual const char inspectorFactoryName Redefined virtual void forEach ForeachFunc f Calls the given function for each object contained Member functions int items Returns the index of last used position 1 void clear As a result the container will be empty Contained objects that were owned by the container will be deleted int add cObject obj Inserts a new object into the array Only the pointer of the object will be stored The return value is the object s index in the array int addAt int m cObject obj Inserts a new object into the array at the given position If the position is occupied the function generates an error message int find cObject obj Searches the array for the pointer of the object passed and returns the index of the first match If the object wasn t found 1 is returned int find const char s Returns the index of the first item in the array that has the name pointed to by s cObject isName is used If no such item was found 1 is returned cObject get int m Returns reference to the mth object in the array or null pointer if the mth position is not used cObject get const char s Returns reference to the first object in the array with name s or null reference NOOBJ if no object with the given name was found cObject operator int m cObject operator
115. SimpleModule 8192 int window_size int n_s N s int nr N x cOutVector eedVector cStdDev eedStats EAs 6 virtual void activity virtual void recalculateTimeout virtual void insertPacketIntoBuffer cMessage packet virtual void resendPacket cMessage packet Pilea Define_Module TransportProtocol void TransportProtocol activity window_size par window_size n_s n_r 0 eedVector setName End to End Delay eedStats setName eedStats Td uer ET ac Note that you may have to use the expanded form of the constructor instead of Module_Class_Members to pass arguments to the constructors of member objects like eedVector and eedStats But most often you dont need to go as far as that for example you can set parameters later from activity as shown in the example above Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 69 To implement another variant of the Transport Protocol which uses a different timeout scheme you could simply subclass TransportProtocol class AdvancedTransportProtocol public TransportProtocol public Module_Class_Members AdvancedTransportProtocol TransportProtocol 8192 virtual void recalculateTimeout Define_Module AdvancedTransportProtocol void AdvancedTransportProtocol recalculateTimeout b boas 5 3 Adding functionali
116. Step e Run e Fast run Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 167 e Express run The running modes have their corresponding buttons on Tkenv s toolbar In Step mode you can execute the simulation event by event In Run mode the simulation runs with all tracing aids on Message animation is active and inspector windows are updated after each event Output messages are displayed in the main window and module output windows You can stop the simulation with the Stop button on the toolbar You can fully interact with the user interface while the simulation is running you can open inspectors etc In Fast mode animation is turned off The inspectors and the message output windows are updated after each 10 events the actual number can be set in OptionslSimulation options and also in the ini file Fast mode is several times faster than the Run mode the speedup can get close to 10 or the configured event count In Express mode the simulation runs at about the same speed as with Cmdenv all tracing disabled Module output is not recorded in the output windows any more You can interact with the simulation only once in a while 1000 events is the default as I recall thus the run time overhead of the user interface is minimal You have to explicitly push the Update inspectors button if you want an update Inspectors In Tkenv objects can be viewed through inspectors To start
117. T 184 11 4 The user interface The source code for the user interface of OMNeT resides in the envir directory common part and in the cmdenv tkenv directories The classes in the user interface are not derived from cOb ject they are completely separated form the simulation kernel 11 4 1 The main function The main function of OMNeT simply sets up the user interface and runs it Actual simulation is done in cEnvir run see later 11 4 2 The cEnvir interface The cEnvir class has only one instance a global object called ev cEnvir ev cEnvir basically is only an interface its member functions hardly contain any code cEnvir maintains a pointer to a dynamically allocated simulation application object derived from TOmnetApp see later which does all actual work cEnvir member functions deal with four basic tasks e V O for module activities actual implementation is different for each user interface e g stdin stdout for Cmdenv windowing in Tkenv e setting up and running the simulation application e provides functions called by simulation kernel objects to get information for example get module parameter settings from the configuration file e provides functions called by simulation kernel objects to notify the user interface of some events This is especially important for windowing user interfaces Tkenv because the events are like this an object was deleted so its inspector window should be close
118. TRANSFORM M no transformation uses bin boundaries previously defined by addBinBound e HIST_TR_AUTO_EPC_DBL automatically creates equiprobable cells e HIST _TR_AUTO_EPC_INT like the above but uses a different hack Creating an object cVarHistogram const char s NULL int numcells 11 int transformtype HIST_TR_AUTO_EPC_DBL Manually adding a cell boundary void addBinBound double x Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 131 Rangemin and rangemax is chosen after collecting the num_firstvals initial observations One cannot add cell boundaries when histogram is already transformed 6 13 3 The k split algorithm Purpose The k split algorithm is an on line distribution estimation method It was designed for on line result collection in simulation programs The method was proposed by Varga and Fakhamzadeh in 1997 The primary advantage of k split is that without having to store the observations it gives a good estimate without requiring a priori information about the distribution including the sample size The k split algorithm can be extended to multi dimensional distributions but here we deal with the one dimensional version only The algorithm The k split algorithm is an adaptive histogram type estimate which maintains a good partitioning by doing cell splits We start out with a histogram range
119. Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 242 Member functions void setup int esiz int siz int delt 5 Clears the current contents and changes element size initial array size and delta parameters void clear Clears the whole contents of the cBag int items Returns the index of last used position 1 int add void data Inserts a new item into the array A copy will be made of the item pointed to by data The return value is the item s index in the array int addAt int pos void data Inserts a new item into the array at the given position int find void data Returns the index of the first item in the array that equals the item pointed to by obj The comparison is done byte by byte If no such item was found 1 is returned void get int m Returns a pointer to the mth item in the array or NULL if the mth position is not used void operator int m The same as get int m With the indexing operator cBag can be used as a vector bool isUsed int m Returns t rue if the mth position exists and an item was inserted there bool remove int m Deletes the mth position in the array Returns true if the mth position was used i e it was actually deleted 15 10 4 Class cWatch Header file carray h sim directory cWatch is an object shell for an ordinary char int long double char or cObject variable in the module functions It
120. Xjo Xni with k equal sized histogram cells with observation counts 7 n2 Nx Each collected observation increments the corresponding observation count When an observation count n reaches a split threshold the cell is split into k smaller equal sized cells with observation counts n i Nj 2 Nix initialized to zero The n observation count is remembered and is called the mother observation count to the newly created cells Further observations may cause cells to be split further e g 1 7 7 1 7 etc thus creating a k order tree of observation counts where leaves contain live counters that are actually incremented by new observations and intermediate nodes contain mother observation counts for their children If an observation falls outside the histogram range the range is extended in a natural manner by inserting new level s at the top of the tree The fundamental parameter to the algorithm is the split factor k Low values of k k 2 and k 3 are to be considered In this paper we examine only the k 2 case Figure 5 7 Illustration of the k split algorithm k 2 The numbers in boxes represent the observation count values For density estimation the total number of observations that fell into each cell of the partition has to be determined For this purpose mother observations in each internal node of the tree must be distributed among its child cells and propagated up to the leaves Let n be the mother observation coun
121. _delayed ici receive pk ici gt decapsulate op_pk_send send msg out gate send msg gate vector index send msg gate_id sendDelayed op_pk_deliver sendDirect op_pk_schedule_self op_ev_canceli scheduleAt simTime timeout msg m cancelEvent msg op_dist op_dist _ load _ outcome To generate random numbers from analytical distributions use uniform intuniform exponential normal truncnormal For custom distributions you can use the histogram classes Histograms can load distribution data from file cDoubleHistogram hist FILE f fopen distribution dat hist loadFromFile f fclose f double rnd hist random Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix A OPNET and OMNeT 195 output vectors output scalars The cOut Vector class can be used cOutVector eed End to end delay double d msg gt creationTime eed record d simTime Output scalar file exists You can write into it with recordScalar recordScalar average delay avg_delay op_topo_parent op_topo_child_ cModule parent parentModule cSubModuleiIterator op_topo_ _assoc_ gate
122. a E a wactohasaasee 19 2 1 1 Hierarchical modules niiiinrisiiimnann n e e e eiia E 19 21 2 Mod le typ S pren a E E R E a aE 19 21 3 Messages gates Tinks 49 irine esans ieiet EE Es 20 2 1 4 Link Character sues os nonen eein E R aE anne 20 2S Parameters eeaeee eR E E EAT S EEEE E E 22 2 1 6 Topology description method sssesessseesseesseessessseressstesstessetsseesseesssseesseesseesse 22 2 2 Programming the alg rithins anoni n n a a A a a ea 23 2 2 1 Creatine Simple mod ler n a a a a E 23 22 2 Object mecha i sms n r odd ee Gi ede Sade AEA GS aah aa ETE a Eh 23 223 Denye neW CLASS 5 on sucastas ua scans e a a a a aa 24 2 2 4 Self describing objects to ease debugging ssesssesssssresseesserssesseessseeessresseesse 24 Dll SIMO OMNE THEa a a cut Ate ae e a kak soc olieas 24 2 3 1 Building and running simulations 0 0 0 0 cece eeeeeecseececeeececeeeeecseeeeeeeesseeeenaeeeenas 24 2 3 2 What is what in the directories so c5issaczentcaactsuasecagukedades Wiedaawenssbsaas iaeastebeaeearaenses 26 3 AN EXAMPLE THE NIM GAME sseseesoeseesoesoesoeecorseeooesoseoecoeroeeooroeecoeeoecoeeoeeoreooreeeooeee 27 JA TOpPOlO Sy merre n a E E E OES 27 32 SUP HEPC S t a a E E E A E E A 29 3 3 Running the simulation sessies iiinis yiii e i nonsense dasdaeca saved evaesdactanetbes 32 DA OME RO KATO S Gon EEE EE EEE EE EEA 33 Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 4 THE NED LANGUAG
123. a number of options in what language to implement the controlling script Some ideas e shell mentioned above e Perl e Tcl e Octave suggested by Richard Lyon see the cont rib octave directory for examples e DOS Win32 batch maybe this is not such a good idea 8 5 User interfaces of simulation executables The user interface is separated from the simulation kernel the two parts interact through a well defined interface This construction makes it possible to implement several types of user interfaces without changing the simulation kernel Also the same simulation model can be executed with different user interfaces without any change in the model files themselves The user would test and debug the simulation with a powerful graphical user interface and finally run it with a simple and fast user interface that supports batch execution User interfaces takes the form of libraries a file or so on UNIX 1ib or d11 file on NT The libraries are interchangeable When the user creates a simulation executable he can pick one of the user interface libraries that he links in Three user interfaces have been implemented e Cmdenv command line user interface for batch execution e Tkenv graphical windowing user interface Tcl Tk The following sections contain more detailed descriptions about each user interface 8 5 1 Cmdenv the command line user interface The command line user interface is a small portable and fast us
124. akemake n The n option means no linking is necessary only compiling has to be done In your toplevel source directory run makemake trafgen router This results in recursive makefiles when you build the simulation make will descend into trafgen and router run make in both then it will link an executable with the object files in the two directories Technical University of Budapest Dept of Telecommunications OMNeT Manual Building Simulation Programs 150 You may need to use the I option if you include files from other directories The I option is for both C and NED files In our example you could run makemake n I router in the trafgen directory and vica versa If youre willing to play with shared and run time loaded libraries several makemake options and the General load libs ini file option leave you enough room to do so 7 2 4 Static vs shared OMNeT system libraries Default linking uses the shared libraries One reason you would want static linking is that debugging the OMNeT class library is more trouble with shared libraries Another reason might be that you want to run the executable on another machine without having to worry about setting LD_LIBRARY_PATH If you want static linking find the build_shared_libs yes line in the configure user script and change it to build_shared_libs no Then you have to re run the configure script and rebuild everything configure
125. ame of cOutVector object number of data single numbers or pairs will be written Columns of a data line vector ID actual simulation time one or two double values Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 244 One can use UNIX tools like sed awk or perl to extract a particular vector etc from the file or and read it in spreadsheets like Excel The cOutVector record member is used to output a value or a value pair This will generate a data line in the file unless the output vector is disabled or the current simulation time is outside a specified interval see member functions Construction destruction copying cOutVector cOutVector amp r Copy constructor cOutVector const char s NULL int tupl 1 Constructor Accepts the object name and the multiplicity of the data it should write to the file at a time Possible values of tuple are 1 or 2 virtual cOutVector Destructor Redefined virtual functions virtual const char className j Returns a pointer to the class name COutVector virtual cObject dup Dupping does not make much sense not implemented virtual void info char buf Redefined Member functions void record double value Records one data to the file It can be used only in the case if the instance of cOut Vector was created with tupl 1 otherwise it gives an error message void record doubl
126. and it builds the model without having to run external programs this means that you can have a single simulation executable that can be used to perform simulation studies on networks with different topologies Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix A OPNET and OMNeT 188 Module parameters Expressions no expressions are allowed arbitrary expressions using only literals or exact copy other parameters of another parameter Parameter passing by value parameters can be passed by value or by reference and be changed during simulation Usage by process models only by process models also to define flexible topologies In OPNET module parameter values can be passed only as is Packet streams or gates Identification Packet streams are Gates are identified by numbered from 0 no names Gate vectors are names can be assigned supported In the code gates can be referenced by ID for greater speed directional Flexible topologies OMNeT not really supported in the NED file parameters can define submodule types count of submodules gates and describe connections Tf really necessary it can be done through C programming writing EMA code and running external program to create a separate model file for each case Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix A OPNET and OMNeT 189 Tracing animation and inte
127. area create new submodule Drag from one submodule to another create new connection Click in empty area switch to select move mode In select move mode Click submodule connection select it Ctrl click submodule conn add to selection Click in empty area clear selection Drag a selected object move selected objects Drag submodule or connection move it Drag either end of connection move that end Drag corner of sub module resize module Drag starting in empty area select enclosed submodules connections Del key delete selected objects Both editing modes Right click on popup menu module submodule connection Double click on submodule go into submodule Click name label edit name Drag amp drop module type from the tree view create a submodule of that type to the canvas Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 61 5 SIMPLE MODULES The activities of simple modules are implemented by the user The algorithms are programmed in C using the OMNeT class library The following sections contain a short introduction to discrete event simulation in general how its concepts are implemented in OMNeT and gives an overview and practical advice on how to design and code simple modules 5 1 Simulation concepts This section contains a very brief introduction into how Dis
128. arsec has no equivalent of OMNeT gates OMNeT s other send functions which send messages through a gate are not present in Parsec Message receiving constructs Parsec entities accept messages with the receive construct Receive has many forms the elegance and power of the Parsec language stems from the receive construct Some illustrative examples receive Request req or receive Release rel or timeout in 5 in timeout with high priority It is possible to add guards to the receive branches receive Request req when regq units lt units or timeout after 5 after timeout with low priority These constructs have to be explicitly programmed in OMNeT using while loops with receive calls and if switch statements in its body The reason OMNeT doesnt have this sort of syntax and functionality is that it is impossible to express with plain C C one cannot avoid the need for a special precompiler Having to use a precompiler however causes some inconvenience during program development and in practice there isnt as much need for this sort of complex receive constructs that would justify making it mandatory to use a precompiler for every source file One may wonder what happens to the messages which have arrived already but have not been accepted by the entity yet because they had no matching receive branch Parsec stores those messages in what it calls the message buff
129. astchildp pointers Contains relationship Only for container classes e g cArray or cQueue Keeping track of contained objects works with another mechanism not the previously mentioned ptrs E g cArray uses a vector cQueue uses a separate list The two mechanisms are independent What cObject does e Owner of a new object can be explicitly given if omitted defaultOwner will be used e An object created through the copy constructor will have the same owner as original and does not dup or take objects owned by the original e Destructor calls free for owned objects see later Rules for derived classes e Objects contained as data members the enclosing object should own them Rules for container objects derived from cObject e they use the functions take obj drop obj free obj e when an object is inserted if takeOwnership is true should take ownership of object by calling take obj takeOwnership defaults to true e when an object is removed they should call drop obj for it if they were the owner e copy constructor copies should dup and take ownership of objects that were owned by the original e destructor doesn t need not call free for objects this will be done in cObject s destructor The class cHead is special case it behaves as a container displaying objects it owns as contents Technical University of Budapest Dept of Telecommunications OMNeT Manual The Design of OMNe
130. at What you will need to have in your project file e your simple module C sources e your NED files e for each NED file the C file it will compile into the _n cc file Place the ned file under the _n cc file in the project tree hierarchy e the OMNeT libraries sim_std lib envir lib plus cmdenv 1lib or tkenv 1lib depending on which user interface you want to link in You also need the Tcl and Tk libraries if you re using Tkenv If you started out from a project file in the distribution you can easily switch from one user interface to another by selecting the Exclude From Parent option for exactly of the omnetpp cmdenv and omnetpp tkenv source pools right click gt Edit Local Options gt Topics Build Attributes gt Exclude from parent If you re going to build a LARGE model be sure to increase the stack size in Options Project options Linkerl32 bit LinkerlReserved stack size The default is 0x1000000 1MB which is hardly enough for OMNeT simulations Increase it to 64MB for example 0x40000000 If the simulation exceeds the stack size configured here you ll get nice exceptions General Protection Faults and the like 7 3 3 Starting from scratch If you want to make your own project files maybe for porting to another compiler here are some hints e Compile as a 32 bit flat console application None of the special libraries OWL MFC Class Library OCF etc are needed e You have to turn off ex
131. at it minimizes that error It has been shown that the cell division error can be reduced to a more than acceptable small value Strictly speaking the k split algorithm is semi online because its needs some observations to set up the initial histogram range However because of the range extension and cell split capabilities the algorithm is not very sensitive to the choice of the initial range so very few observations are enough for range estimation say Npre 10 Thus we can regard k split as an on line method K split can also be used in semi online mode when the algorithm is only used to create an optimal partition from a larger number of Npre observations When the partition has been created the observation counts are cleared and the Npre observations are fed into k split once again This way all mother non leaf observation counts will be zero and the cell division error is eliminated It has been shown that the partition created by k split can be better than both the equi distant and the equal frequency partition OMNeT contains an experimental implementation of the k split algorithm the cKSp1lit class Research on k split is still under way The cKSplit class TBD comments Member functions void setCritFunc KSplitCritFune _critfunc double _critdata Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 133 void setDivFunc KSplitDivFunc _divfunc double _divdata voi
132. ay strings can be assigned to modules submodules and gates a connection s display string is stored in its from gate Display strings can be hardcoded into the NED file or specified in the configuration file Hardcoded display strings take precedence over the ones given in ini files Format of display string are documented in the User Interfaces chapter Display strings can appear in the DisplayStrings section of the ini file They are expected as entries in one of the following forms moduletype moduletype submodulename moduletype inputgatename moduletype submodulename outputgatename As with parameter names wildcards are allowed in module types submodule and gate names Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 160 8 2 7 Specifying seed values As is was pointed out earlier it is of great importance that different simulation runs and different random number sources within one simulation run use non overlapping sequences of random numbers In OMNeT you have three choices 1 Automatic seed selection 2 Specify seeds in the ini file with the help of the seedtool program see later 3 Manually set the seed from within the program If you decide for automatic seed selection do not specify any seed value in the ini file For the random number generators OMNeT will automatically select seeds that are 1 000 000 values apart in
133. b Construction destruction copying cQueue cQueue amp queue Copy constructor Contained objects that are owned by the queue will be duplicated so that the new queue will have its own copy of them cQueue const char name NULL CompareFunc cmp NULL bool a false Constructor It accepts the object name the address of the comparing function the sorting order ascending t rue descending false and the list the object will join virtual cQueue Destructor Deletes all contained objects that were owned by it virtual cObject dup cQueue amp operator cQueue amp queue Duplication and assignment work all right with cQueue Contained objects that are owned by the queue will be duplicated so that the new queue will have its own copy of them Redefined virtual functions virtual const char className j Returns pointer to the class name string cQueue virtual void info char buf virtual const char inspectorFactoryName Redefined virtual void forEach ForeachFunc f Calls the given function for each object contained Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 237 Member functions void setup CompareFunc cmp NULL bool a false Changes the sort function and the sorting order Doesn t re sort the contents of the queue void clear As a result the container will be empty Contained objects that
134. bers SlidingWindow cSimpleModule 8192 virtual void activity module type registration Define_Module SlidingWindow implementation of the module class void SlidingWindow activity int window_size par window_size In order to be able to refer to this simple module type in NED files we should have an associated NED declaration which might look like this file swp ned simple SlidingWindow parameters window_size numeric const gates in from_net from_user out to_net to_user endsimple 5 2 2 The module declaration The module declaration e announces that you re going to use the class as a simple module type e associates the module class with an interface declared in NED Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 65 Forms of module declaration Module declarations can take two forms Define Module classname Define Module _Like classname neddeclname The first form associates the class subclassed from cSimpleModule with the NED simple module declaration of the same name For example the Declare_Module SlidingWindow line would ensure that when you create an instance of SlidingWindow in your NED files the module has the parameters and gates given in the simple SlidingWindow NED declaration and the implementation will be an instance of the SLidingWindow C class The second fo
135. bject amp operator int Steps to the next previous object in the queue 15 9 4 Inline functions The following inline functions are for finding objects by name inline cNetworkType findNetwork const char s j Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 238 Equals cNetworkType networks find s inline cModuleType findModuleType const char s Equals cModuleType modtypes find s inline cLinkType findLink const char s Equals cLinkType linktypes find s inline cFunctionType findFunction const char s j Equals cFunctionType functions find s 15 9 5 class cHead cHead is the head of a cObject chain cObject and its derived classes contain pointers that enable the objects to be a node in a double linked list cObject has member functions to link and unlink to and from double linked lists See documentation on cObject setOwner The head of such lists is always a cHead object The lists are a means that provide that each object in the system is part of an object tree and can be accessed through pointers starting from a given point The existence of such hierarchy is necessary for a user interface where we want each object to be visible to the user It is also unavoidable when we want the simulation to be restartable we need to destroy objects created by the running simulation to start a new one Last it enables that all ob
136. ble The use of this macro allows the function to be used in expressions inside NED network descriptions The commonly used lt math h gt functions have Define _Function lines in the OMNeT simulation kernel sources Register_Class CLASSNAME Registers a class for use with the createOne function cObject createOne const char classname This function creates an object of the type given by the string argument createOne is used by the distributed execution part of OMNeT for unpacking objects from a buffer that was sent over from another machine processor Each class that should be able to travel between machines processors should be registered with the Register_Class macro in one of the C source files The Register_Class macro expands to the definition of acClassRegister object Interface CLASSNAME Gate NAME TYPE Parameter NAME TYPES Machine NAME End Simple module declaration macro Gate types can be Input or Output parameter types can be Anytype or Numeric Module_Class_Members CLASSNAME BASECLASS STACKSIZE This macro facilitates the declaration of a simple module type s class The macro is used like this class CLASSNAME public cSimpleModule Module_Class_Members CLASSNAME cSimpleModule 8192 virtual void activity hi The macro expands to the definition of member functions which the user does no
137. by PVM to describe what computers will participate in the virtual machine where the executables in our case the OMNeT programs are located on each computer what working directories should be set etc It is advisable to have a common shared directory mounted on all participating hosts this eliminates the tedious work of having to copy files to all hosts again and again If using OMNeT it is a good idea to write separate pvmhosts files for each simulation program Since simulation programs are typically in separate directories the pvmhosts file in each directory can name that directory as executables directory and working directory for each host This way there is no need to create soft links or explicitly name directories in the OMNeT ini files Technical University of Budapest Dept of Telecommunications OMNeT Manual Parallel Execution 177 Each line in the pvmhosts file describes one host An example line this all should be a single line whale ip whale hit bme hu lo andras home andras pvm pvm ep home andras omnetpp projects fddi wd home andras omnetpp projects fddi To start PVM with this configuration cd omnetpp projects fddi pvm pvmhosts 10 1 4 Mapping logical machines to physical ones The on phrases in the NED descriptions specify the logical machine s on which the module is run The machine parameters are mapped to physical machines in the Machines section of the configuration file file
138. c function 6 12 7 Summary The various cPar types and the member functions used to manipulate them are summarized in the following table Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library Type Type Member functions Description char name l l l i f P void pointer setStringVal const char stringVal op const char op const char setBoolValue bool bool boolValue op bool op bool tLongValue long ng longVal long Lue const char lue lue lue double leValue double doubleValue op double setDoubleVal ue sXElem int double doubl leValue op double tDoubleVal ue uble doub cStatistic leValue double setPointerValue void void pointerValue op void op void string value Short strings len lt 27 are stored inside cPar object without using heap allocation boolean value Can also be retrieved from the object as long 0 or 1 signed long integer value Can also be retrieved from the object as double double precision floating point value Mathematical function with constant arguments The function is given by its pointer it must take 0 1 2 or 3 doubles and return a double This type is mainly used to generate rand
139. cannot safely delete a compound module from a simple module in it you must delegate the job to a module outside the compound module Creating connections There are two functions that you can use to connect gates For a normal user they are useful for creating connections to dynamically created modules cModule src_module int src_gatenumber cLinkType channeltype cModule dest_module int dest_gatenumber connect connect cModule src_module int src_gatenumber cPar delay cPar error cPar datarate cModule dest_module int dest_gatenumber Any of the channeltype delay error and datarate pointers can be NULL An example connect this findGate out cLinkType NULL module module gt findGate in 0 6 8 Routing support cTopology 6 8 1 Overview The cTopology class was designed primarily to support routing in telecommunication or multiprocessor networks A cTopology object stores an abstract representation of the network in graph form e each cTopology node corresponds to a module simple or compound and e each cTopology edge corresponds to a link or series of connecting links Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 113 You can specify which modules either simple or compound you want to include in the graph The graph will include all connections among the selected modules In the graph all node
140. ce parameters interarrival_time num_of_messages const address string gates endsimple If the parameter type is omitted numeric is assumed Practically this means that you only need to explicitly specify the type for string bool or char valued parameters Note that the actual parameter values are given later when the module is used as a building block of a compound module type or as a system module When the user writes the word const before the parameter it is converted to constant that is the parameter s value is replaced by its evaluation This can be important when the original value was a random number or an expression One is advised to write out the const keyword for each parameter that should be constant Beware when using const and by reference parameter passing ref modifier see later at the same time Converting the parameter to constant can affect other modules and cause errors that are difficult to discover 4 4 2 Simple module gates Gates are the connection points of modules The starting and ending points of the connections between modules are gates OMNeT supports simplex one directional connections so there are two kinds of gates input and output Messages are sent through output gates and received through input gates Gates are identified with their names Gate vectors are supported a gate vector contains a number of single gates Gates are declared by listing their names in the gates
141. ception handling it conflicts with the coroutine library somehow In the IDE OptionslProject gt C Options gt Exception Handling RTTI gt clear Enable exceptions It must be done both when compiling the libraries and when compiling simulation applications e Borland C does not recognize the cc extension as C You have to teach it Options Tools gt select CppCompile gt Edit gt Advanced gt add the cc extension to the Translate From and Default For entries Do the same with the EditText tool e You also have to teach Borland C how to handle ned files Select OptionsITools gt New Fill in the dialog as follows Name NEDCompile Path src nedc nedc exe Command Line NOSWAP CAP MSG BORL2MSG EDNAME Menu Text NED Compile Help Hint OMNeT NED compiler Select Advanced and fill in the dialog Technical University of Budapest Dept of Telecommunications OMNeT Manual Building Simulation Programs 152 Type Translator Translate From ned Translate To cc Default For ned 7 4 Using Win32 with MSVC 7 4 1 Prerequisite install Tcl Tk Download and install Tcl Tk You need at least version 8 0p1 but it s better to download the latest version currently 8 2 7 4 2 Installing OMNeT Download the omnetpp source code and documentation from the omnetpp site If you downloaded a zip file extract its contents to the installation directory If you downloaded a tgz file copy it to
142. choose InspectINetwork from the menu Usage should be obvious just use double clicks and popup menus that are brought up by right clicking In Step Run and Fast Run modes inspectors are updated automatically as the simulation progresses To make ordinary variables int double char etc appear in Tkenv use the WATCH macro in the C code In list dialogs entries begin with text like ptr0x8000ab7e Yes it is really the object pointer knowing it is extremely useful if you re running the simulation under a debugger such as gdb Configuring Tkenv In case of nonstandard installation it may be necessary to set the OMNETPP_TKENV_DIR environment variable so that Tkenv can find its parts written in Tcl script The default path from where the icons are loaded can be changed with the OMNETPP_BITMAP_PATH variable which is a semicolon separated list of directories and defaults to omnetpp dir bitmaps bitmaps The ini file options accepted by Tkenv are Tkenv use mainwindow yes print banners yes breakpoints enabled yes update freq fast 10 update freq express 500 animation delay 0 3s The above options can also be set from within Tkenv itself from a configuration dialog box Embedding TCL code into the executable A significant part of Tkenv is written in TCL in several tcl script files The default location of the scripts is passed compile time to tkapp cc and it can be over
143. cisasseenvacessedtenceronvevsactionsdacayeysssevsdhedannsncdnsestenadees A NetWork ICE HIN MOM is sucuter ei ies n eee tae shee eal eadea setae Aaah tes 4 8 Support for parallel execution ii sssccisssisaetisesiacsiacsesaansvaeadsvecgccevanscsanea lessdccansaeens 4 8 1 Extensions to the compound module and system definitions 4 8 2 Conditional On SCCtIONS ccccccseeseesececcccceseessseececccessssuseaesesceesseees Bh Oe TR OTE STO MIG NRE E EEEE A E euros asus natde gcty E EE 4 9 1 Using parameters in expressions ref and ancestOr ccceeeeeeeeeeees Bh OO MCT ALON cede ieee CE ES EN EES Le BGs is ed a a 4 9 3 The sizeof and index Operators eeeceescecessceceeececeeececeeceeceseeeesaeees AOA Time Constants A cede ees Got Gis ieee mee Aaa hie ote 4 9 5 Randon VAC Snein cca veda a EE E ERa AOSA DATON VAG E EE E ENE TA as Tt ADT FONCIONS e a E E E a E a Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 5 4 9 3 Display SINS Sarn i aren eucynuie a atasateo tees ation aaa 56 4 10 GNED Graphical NED Editor s c22 i esger ee Gece Bd eee 58 Ss SIMPLE MODULES sicavesssedsscsssscevesesttecessdecanedavebivanesaccuacepuieisabecesnascatetasseacaeesenssscestodss 61 Sol SUMUlAMOM CONCEDIS 422s fatal ie eae tele ee elie ee le tae 61 Del IsCrele Venti IU ALON 25 aoa eter E E wise ean van eure cn ane 61 Del 2 The eyent WOOP testes erate iti peat nd ele Ne Sends ee 61 5 1 3
144. ck gatesizeblock on_block Technical University of Budapest Dept of Telecommunications 207 OMNeT Manual Appendix C NED Language Grammar 208 ON IF expression J on_machine substparamblock PARAMETERS IF expression substparamname substparamvalue substparamvalue ANCESTOR REF name parexpression gatesizeblock GATESIZES IF expression gatename vector connblock CONNECTIONS NOCHECK connection connection normalconnection loopconnection loopconnection FOR index DO normalconnection ENDFOR index indexvariable expression expression normalconnection gate gt lt gate IF expression gate gt channel gt gate IF expression gate lt channel lt gate IF expression channel channeltype DELAY expression ERROR expression DATARATE expression gate i modulename vector gatename vector networkdefinition NETWORK networkname moduletype on_block substparamblock ENDNETWORK vector expression parexpression expression otherconstvalue expression expression expression expression expression expression expression expression expression Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix C NED Lan
145. class MySimpleModule variables for statistics collection activity finish J MySimpleModule activity declare local vars and initialize them initialize statistics collection variables while true MySimpleModule finish record Statistics into file Advantages and drawbacks Advantages e initialize not needed state can be stored in local variables of activity e process style description is a natural programming model in many cases Drawbacks e memory overhead stack allocation may unacceptably increase the memory requirements of the simulation program if you have several thousands or ten thousands of simple modules e run time overhead switching between coroutines is somewhat slower than a simple function call Other simulators Coroutines are used by a number of other simulation packages e All simulation software which inherit from SIMULA e g C SIM are based on coroutines although all in all the programming model is quite different e The simulation parallel programming language Maisie and its successor Parsec from UCLA also use coroutines although implemented on with normal preemptive threads The philosophy is quite similar to OMNeT Parsec being just a programming language has a more elegant syntax but much less features than OMNeT e Many Java based simulation libraries are based on Java threads Technical University of Budapest Dept of Telecommunications
146. classes are there to generate random numbers from equidistant cell or equiprobable cell histograms This feature is documented later with the statistical classes 6 10 Container classes 6 10 1 Queue class cQueue Basic usage cQueue is a container class that acts as a queue cQueue can hold objects of type derived from cObject almost all classes from the OMNeT library such as cMessage cPar etc Internally cQueue uses a double linked list to store the elements As an example of use the simple modules put aside queues pbutAsideQueue member are cQueues which store cMessage objects However the Future Event Set FES is not a cQueue it is implemented with heap class cMessageHeap because it is a lot more efficient A queue object has a head and a tail Normally new elements are inserted at its head and elements are removed at its tail HEAD TAIL oes removal insert pop Fig 5 2 What is what with cQueue The basic cQueue member functions dealing with insertion and removal are insert and pop They are used like this cQueue queue my queue cMessage msg insert messages for int i 0 i lt 10 i msg new cMessage queue insert msg Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 119 remove messages while queue empty msg cMessage queue pop delete msg The length member
147. co com Technology tcl Tcl html Brief tutorial http nacphy physics orst edu DATAVIS datavis html Reference http www cm cf ac uk Latex Gnuplot gnuplot html PlotMTV http cauchy math edu workshop Plotmtv plotmtv html Brief tutorial http nacphy physics orst edu DATAVIS xmgr html Technical University of Budapest Dept of Telecommunications
148. copying The user normally doesnt create new modules directly so these member functions are not documented here Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 219 Redefined virtual functions virtual const char className Returns pointer to the class name string cModule virtual const char fullName Returns full name of the module in a static buffer in the form name or name index virtual const char fullPath int 1 Returns full path from the system module to the module in a static buffer in the form system_module submodule 1 subsubmodule name index virtual void info char buf virtual const char inspectorFactoryName Redefined Member functions bool isSimple 0 Pure virtual function It is redefined in cSimpleModule to return true and in cCompoundModule to return false bool isOnLocalMachine Used with parallel execution determines if the module is on the local machine See the user manual for more info int id Returns the index of the module in the module vector cSimulation simulation cModule parentModule Returns reference to the module s parent int index Returns the index of the module if it is multiple otherwise 0 int size Returns the size of the multiple module or 1 Module gates int gates Returns total number of module gates cGate amp ga
149. cording statistics into file cOut Vector class e collecting simple statistics cSt dDev and cWeightedStddev classes e distribution estimation cLongHistogram cDoubleHistogram cVarHistogram cPSquare cKSplit classes e making variables inspectable in the graphical user interface Tkenv the WATCH macro cWatch class e sending debug output to and prompting for user input in the graphical user interface Tkenv the ev object cEnvir class 6 1 Class library conventions Base class Classes in the OMNeT simulation library are derived from cObject Functionality and conventions that come from cObject e name attribute e className member and other member functions giving textual information about the object e conventions for assignment copying duplicating the object e ownership control for containers derived from cOb ject e support for traversing the object tree e support for inspecting the object in graphical user interfaces Tkenv e support for automatic cleanup garbage collection at the end of the simulation Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 93 Classes inherit and redefine several cObject member functions in the following well discuss some of the practically important ones Setting and getting attributes Member functions that set and query object attributes follow consistent naming The setter member function has the form setSomethi
150. crete Event Simulation DES works in order to introduce terms well use when explaining OMNeT concepts and implementation If you re familiar with DES you can skip this section 5 1 1 Discrete Event Simulation A Discrete Event System is a system where state changes events happen at discrete points of time and events take zero time to happen It is assumed that nothing i e nothing interesting happens between two consecutive events that is no state change takes place in the system between the events in contrast to continuous systems where state changes are continuous Those systems that can be viewed as Discrete Event Systems can be modeled using Discrete Event Simulation Continuous systems are modelled using differential equations and suchlike For example computer networks are usually viewed as discrete event systems Some of the events are e start of a packet transmission e end of a packet transmission e expiry of a retransmission timeout This implies that between two events such as start of a packet transmission and end of a packet transmission nothing interesting happens That is the packet s state remains being transmitted Note that the definition of events and states always depends on the intent and purposes of the person doing the modeling If we were interested in the transmission of individual bits we would have included something like start of bit transmission and end of bit transmission among our even
151. ct can fully work together with other objects 2 2 4 Self describing objects to ease debugging The class library is designed so that objects can give textual information about themselves This makes it possible to peek into a running simulation program through an appropriate user interface one can examine and modify the internal data structures of a running simulation This feature helps the user to get some insight what is happening inside the model and get hands on experience A unique feature called snapshot allows the user to dump the contents of the simulation model or a part of it into a text file The file will contain textual reports about every object this can be of invaluable help at times of debugging Ordinary variables can also be made to appear in the snapshot file Snapshot creations can be scheduled from within the simulation program or done from the user interface 2 3 Using OMNeT 2 3 1 Building and running simulations This section gives some idea how to work with OMNeT in practice issues like model files compiling and running simulations are discussed An OMNeT model consists of the following parts e NED language topology description s which describe the module structure with parameters gates etc They are files with ned suffix NED files can be written with any text editor or using the GNED graphical editor e Simple modules sources They are C files with h cc suffix The simulation system provides
152. ctions virtual const char className Returns pointer to the class name string cSimpleModule virtual void info char buf virtual const char inspectorFactoryName Redefined bool isSimple Returns true virtual void scheduleStart simtime_t t Creates a starting message for the module virtual void deleteModule Deletes a dynamically created module Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 222 Public data members cQueue putAsideQueue Represents the put aside queue where the simulator puts unexpected messages that arrive to the module The putaside queue is configured to sort the messages by arrival time and priority like the event queue does You can freely use the contents of the putaside queue the simulator kernel does not do anything with it it only inserts messages See class cQueue and the functions cSimpleModule wait cSimpleModule receiveNewOn cSimpleModule receive for more information cHead locals List of local variables of module function You do not normally need to care about this object cHead members Data members of derived classes If you derive a class from cSimpleModule and add object members you should call object gt setOwner members foreach object you add The module function virtual void activity Contains the module function It is empty in cSimpleModule a
153. ctions of the classes are documented Some of the classes which are not visible to the user have been omitted and of the classes that are documented some member functions are very briefly or not at all described However I hope this material can be used as a handy reference 15 2 Constants types simple functions Header file defs h and util h sim directory 15 2 1 Constants types NO cXX NOOBJ The NO classname macro can be used for NULL pointers of a specific type For most classes in the simulation library there is also a one word shorthand for NO classname NOOBJ NOPAR NOMSG etc simtime_t MAXTIME Type to measure model time defined to double its maximum value The value of m 6 14159 15 2 2 Function typedefs typedef int CompareFunc cObject cObject Type of function used in a cQueue object to compare items See documentation on cQueue for more information typedef bool ForeachFunc cObject bool Type of function to be passed to forEach See cObject forEach for more information typedef doubl typedef doubl MathFunc MathFuncNoArg typedef doubl MathFunclArg double typedef doubl MathFunc2Args double double typedef double MathFunc3Args double double double Used by cPar and sXElem ooo ao typedef void VoidDelFunc void typedef void VoidDupFunc void Memory management functions for void
154. d a message was sent so it can be displayed a breakpoint was hit 11 4 3 Implementation of the user interface simulation applications The base class for simulation application is TOmnet App Specific user interfaces such as TCmdenv TOmnet TkApp are derived from TOmnetApp TOmnetApps member functions are almost all virtual e Some of them implement the cEnvir functions described in the previous section e Others implement the common part of all user interfaces for example reading options from the configuration files making the options effective within the simulation kernel e The run function is pure virtual it is different for each user interface TOmnetApp s data members e a pointer to the object holding configuration file contents type cInifile Technical University of Budapest Dept of Telecommunications OMNeT Manual The Design of OMNeT 185 e the options and switches that can be set from the configuration file these members begin with opt_ Concrete simulation applications e add new configuration options e provide a run function e implement functions left empty in TOmnetApp like breakpointHit objectDeleted 11 5 Writing inspectors for TkEnv TBD Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix A OPNET and OMNeT 186 12 APPENDIX A OPNET AND OMNET 12 1 Comparison of OPNET and OMNeT OPNET from MIL3 Inc is a state o
155. d equal if they are very close if they differ less than simtime_precision However it is not at all clear how small simtime_precision should be the mechanism incurs some run time overhead and all in all I m not sure the whole thing would be of more benefit than trouble 5 1 5 FES implementation The implementation of the FES is a crucial factor in the performance of a discrete event simulator In OMNeT the FES is implemented with binary heap the most widely used data structure for this purpose Heap is also the best algorithm we know although exotic data structures like skiplist may perform better than heap in some cases In case youre interested the FES implementation is in the cMessageHeap class but as a simulation programmer you won t ever need to care about it Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 64 5 2 Defining simple module types 5 2 1 Overview The C implementation of a simple module consists of e declaration of the module class your class subclassed from cSimpleModul1e either directly or indirectly e amodule type registration De fine_Module or Define_Module_Like macro e implementation of the module class For example the C source for a Sliding Window Protocol implementation might look like this file swp cc include lt omnetpp h gt module class declaration class SlidingWindow public cSimpleModule Module_Class_Mem
156. d most of the time There was no mention of tools that could generate the driver entity based on some higher level description OMNeT compound modules have no equivalent in PARSEC All entities are at the same level there s no way to express hierarchy PARSEC has no notion of module gates and there are no connections in the OMNeT sense among the entities This means that when sending messages the receiving entity must be explicitly named Since the program contains no explicit topology information an entity initially has no information about its communication partners it knows no enames except its own The usual practice is that the driver entity sends the necessary enames in an initialization message to each entity For illustration see the add_to_your_sorc message type from the above code fragment The message name itself is quite descriptive The consequence of the lack of compound modules and module gates is that it is a complicated and tricky task to set up networks with but the most trivial topology It is also very difficult to write reusable simulation components without well defined interfaces and structuring compound modules Problems with splitting up the entity body During programming the code of an entity may become so large that it is no longer feasible to keep it within a single function body In OMNeT you can solve the problem by distributing the simple module class s activity code into new member func
157. d overflows are returned by the underflowCell and overflowCell member functions HHH underflows cells overflows Fig 5 5 Histogram structure after setting up the cells You create a P object by specifying the number of cells cPSquare psquare interarrival times 20 Afterwards a cPSquare can be used with the same member functions as a histogram There are three member functions to explicitly return cell boundaries and the number of observations is each cell cells returns the number of cells basepoint int k returns the kth base point cell int k returns the number of observations in cell k and cellPDF int k returns the PDF value in the cell These functions work for all histogram types cPSquare and cKSplit cell 0 1 base point 0 1 Fig 5 6 base points and cells An example long n histogram samples for int i 0 i lt histogram cells i double cellWidth histogram basepoint i 1 histogram basepoint i int count histogram cell i double pdf histogram cellPDF i Ts cx The pdf x and cdf x member functions return the value of the probability density function and the cumulated density function at a given x respectively The random member function generates random numbers from the distribution stored by the object Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 130
158. d rangeExtension bool enabled struct sGrid int parent index of parent grid int reldepth depth reldepth rootgrid s reldepth long total sum of cells amp all subgrids includes mother int mother observations inherited from mother cell int cells K cell values int treeDepth int treeDepth sGrid amp grid double realCellValue sGrid amp grid int cell void printGrids sGrid amp grid int k sGrid amp rootGrid 6 13 4 Transient detection and result accuracy In many simulations only the steady state performance i e the performance after the system has reached a stable state is of interest The initial part of the simulation is called the transient period After the model has entered steady state simulation must proceed until enough statistical data have been collected to compute result with the required accuracy Detection of the end of the transient period and a certain result accuracy is supported by OMNeT The user can attach transient detection and result accuracy objects to a result object cStatistic s descendants The transient detection and result accuracy objects will do the specific algorithms on the data fed into the result object and tell if the transient period is over or the result accuracy has been reached The base classes for classes implementing specific transient detection and result accuracy detection algorithms are
159. diate true TBD const char displayString int type TBD Warning messages bool warnings Warning messages can be enabled disabled individually for each module This function returns the warning status for this module t rue enabled fal se disabled void setWarnings bool wr Enables or disables warnings for this module t rue enable false disable Other functions bool checkInternalConnections For compound modules it checks if all gates are connected inside the module it returns t rue if they are OK for simple modules it returns t rue virtual void scheduleStart simtime_t t 0 Pure virtual function it is redefined in both cCompoundModule and cSimpleModule It creates starting message for dynamically created module or recursively for its submodules See the user manual for explanation how to use dynamically created modules virtual void deleteModule 0 Pure virtual function it is redefined in both cCompoundModule and cSimpleModule Deletes a dynamically created module and recursively all its submodules int findSubmodule const char submodname int idx 1 Finds an immediate submodule with the given name and optional index and returns its module ID If the submodule was not found returns 1 Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 221 cModule submodule const char submodname int idx 1 Finds an immedia
160. difier has a significant role Consider the following example simple Sender parameters num_of_outgates gates out outgate num_of_outgates endsimple Sender Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 48 simple Receiver gates in ingate endsimple Receiver module Network parameters num_of_mods const submodules sender Sender parameters num_of_outgates num_of_mods receiver Receiver num_of_mods connections for i 1 num_of_mods do sender outgate i gt receiver i ingate endfor endmodule network net Network parameters num_of_mods normal 5 2 endnetwork If parameter num_of_mods wasn tconst the following would happen normal 5 2 would be substituted for the num_of_mods There are three places where an evaluation of num_of_mods that is normal 5 2 is done they are typed in italics in the example It is likely that these evaluations would not result in the same value and consequently the gate vector sizes would not match each other and the end value of the for statement Thus the loop connection would not be created properly Using const for the parameter num_of_mods prevents this from happening an evaluation of normal 5 2 is substituted for num_of_mods and an equal number of gates are created 4 6 3 Design patterns for compound modules Several approaches can be used when you want to create complex topologies which
161. dleMessage cMessage msg e void finish In the initialization step OMNeT builds the network it creates the necessary simple and compound modules and connects them according to the NED definitions OMNeT also calls the initialize functions of all modules The activity and handleMessage functions are called during event processing This means that the user will implement the model s behavior in these functions Activity and handleMessage implement different event processing strategies for each simple module the user has to redefine exactly one of these functions activity is a coroutine based solution Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 63 which implements the process interaction approach coroutines are non preemptive cooperative threads and handleMessage is a function called for each event Modules written with these functions can be freely mixed within a simulation model so you can choose per module basis The finish functions are called when the simulation terminates successfully It is the place of writing statistics All these functions will be discussed later in detail 5 1 4 Events in OMNeT OMNeT uses messages to represent events Each event is represented by an instance of the cMessage class or one its subclasses there is no separate event class Messages are sent from one module to another this means that the place where the
162. dule hierarchy nesesssessessseesseeesseeesseessesssesssetessseesseessensseeesseeesseesseesse 109 6 7 Dynamic module creation sivsisyscsevi se sesssacvevascisdseateaspceeavassesnetsunasessaadgendeadpavaccts seneeaseaeantisevs 110 6 8 Routing support cTopology sessseseesssesseessessseeesseetsseesserssesseeessetessresseesseeesseeesseesseesse 112 OS OVERVIEW A EE A E EL 112 6 9 2 Basic Usage era a E N a a a te dea oes E N 113 G25 UOILESU Pall S25 aes E E E T E EE 114 6 9 Generating random numbers 5 co c25 eres saydeesdeeasIisscaaendaguactswedecas shedhans duel aawdsacetaseaeanasedeeesded 116 6 9 1 Using random number generators directly 00 0 eeeceeeeceecesececsteeeeeteeeesteeeenaeeees 116 Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 7 6 9 2 Random numbers from distributions ceecceeeecceceeececeeeeeeeteeeeeteeeeseeeenaeeees 117 6 9 3 Random numbers from histograms eccceeescecsececeeeeeceeeeeeseeeeeneeeeseeeenaeeees 118 G LO Container CLASSES Gs n gate aot ta tines a yee sade tect ads aa atc eos 118 6 10 1 Quevic Class CQUCUE insane e aa i r E E A 118 6 10 2 Expandable atray CAIray acu dient neds aintndn pe oad ea ee 120 6 11 Non object container classes v5 cis sicisinissesesssonian es deesaanasdesaacvunacceds seasaasd lac decestacadeuansdactecesd 121 6 12 The parameter class CP atc Goa essen SG ee os ire OI E 121 6 2 NASI WSOC as cat oor a a en csaad E de
163. e Returns pointer to the current phase string void pause const char s If the user interface supports step by step execution one can stop execution at each receive call of the module function and examine the objects variables etc If the state of simulation between receive calls is also of interest one can use pause calls The string argument if given sets the phase string so pause here is the same as setPhase here pause void breakpoint const char label Specifies a breakpoint During simulation if execution gets to a breakpoint call and breakpoints are active etc the simulation will be stopped if the user interface can handle breakpoints Non typed heap allocation deallocation for module functions void memAlloc size_t m Dynamic memory allocation This function should be used instead of the global malloc from inside the module function activity if deallocation by the simple module constructor is not provided Dynamic allocations are discouraged in general unless you put the pointer into the class declaration of the simple module class and provide a proper destructor Or you can use container classes cArray cQueue void memFree void amp p Frees a memory block reserved with the malloc described above and NULLs the pointer Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 225 15 6 Message classe
164. e 1 seems like we going to lose ev lt lt Taking lt lt move some debug output The ev lt lt out of lt lt num_sticks object represents the user lt lt sticks n interface of the simulator cMessage msgout new cMessage create empty message msgout gt setKind move use message kind as storage for move send msgout out send the message to Game The Game module first waits for a message from both players and extracts the message names that are also the players names Then it enters a loop with the player_to_move variable alternating between 1 and 2 With each iteration it sends out a message with the current number of sticks to the corresponding player and gets back the number of sticks taken by that player When the sticks are out the module announces the winner and ends the simulation The source file game cc df part of NIM an OMNeT demo simulation i include lt stdio h gt include lt string h gt include omnetpp h derive Game from cSimpleModule class Game public cSimpleModule Module_Class_Members Game cSimpleModule 8192 this is a macro it expands to constructor definition etc Technical University of Budapest Dept of Telecommunications register the simple module class to OMNeT OMNeT Manual An Example The NIM Game 31 8192 is the size for the coroutine stack in bytes
165. e just go one level up in the directories http www hit bme hu phd vargaa Because k split is not very well known we ll devote a section to it Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 128 Basic usage One can insert an observation into a statistic object with the collect function or the operator they are equivalent cSt dDev has the following methods for getting statistics out of the object samples min max mean stddev variance sum sqrSum with the obvious meanings An example usage for cSt dDev cStdDev stat stat for int i 0 i lt 10 i stat collect normal 0 1 long num_samples stat samples double smallest stat min largest stat max double mean stat mean standard_deviation stat stddev variance stat variance 6 13 2 Distribution estimation Initialization and usage The distribution estimation classes the histogram classes cPSquare and cKSplit are derived from cDensityEstBase Distribution estimation classes except for cPSquare assume that the observations are within a range You may specify the range explicitly based on some a priori info about the distribution or you may let the object collect the first few observations and determine the range from them Methods which let you specify range settings are part of cDensityEstBase The following member functions exist
166. e nim ned Nim compound module system module import nim_mod module Nim submodules game Game parameters num_sticks intuniform 21 31 first_move intuniform l 2 playerl SmartPlayer Technical University of Budapest Dept of Telecommunications OMNeT Manual An Example The NIM Game 29 player2 SimplePlayer connections playerl out gt game from_playerl playerl in lt game to_playerl player2 out gt game from_player2 player2 in lt game to_player2 endmodule system module creation network nim Nim endnetwork 3 2 Simple modules The module types SmartPlayer SimplePlayer and Game are implemented in C using the OMNeT library classes and functions Each simple module type is derived from the C class cSimpleModule with its activity member function redefined The activity functions of all simple modules in the network are executed as coroutines so they appear as if they were running in parallel Messages are instances of the class cMessage We present here the C sources of the SmartPlayer and Game module types The SmartPlayer first introduces himself by sending its name to the Game module Then it enters an infinite loop with each iteration it receives a message from Game with the number of sticks left it calculates its move and sends back a message containing the move Here s the source include lt stdio h gt
167. e separated by special lines 6 15 Deriving new classes Nearly all classes in the simulation class library are descendants of cOb ject If you want to derive a new class from cObject or a cObject descendant you must redefine some member functions so that objects of the new type can fully co operate with other parts of the simulation system A more or less complete list of these functions are presented here Do not be embarrassed at the length of the list most functions are not absolutely necessary to implement For example you do not need to redefine forEach unless your class is a container class e default constructor copy constructor The copy constructor can simply call the assignment operator e operator the assignment operator copies object contents from another object e dup duplicates the object by creating an exact copy uses copy constructor Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 136 e className returns class name string e info returns a one line info about object contents e writeContents write a more detailed report about the object into a file e forEach iterates through all contained objects if any e netPack netUnpack they are needed only if objects of this type will be sent over PVM from one segment to another The net Pack netUnpack functions of the library classes are in the sim pvm pvmpack cc file e inspector
168. e a given output gate leads finally you would have to walk along the path like this the ownerModule member function returns the module to which the gate belongs cGate gate gate out while gate gt toGate NULL gate gate gt toGate cModule destmod gate gt ownerModule but luckily there are two convenience functions which do that sourceGate and destinationGate Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 109 6 6 Walking the module hierarchy Module vectors If a module is part of a module vector the index and size member functions can be used to query its index and the vector size ev lt lt This is module lt lt module gt index lt lt in a vector of size lt lt module gt size lt lt n Module IDs Each module in the network has a unique ID that is returned by the id member function The module ID is used internally by the simulation kernel to identify modules int myModulelId id If you know the module ID you can ask the simulation object a global variable to get back the module pointer int id 100 cModule mod simulation module id Module IDs are guaranteed to be unique even when modules are created and destroyed dynamically That is an ID which once belonged to a module which was deleted is never issued to another module later Walking up and down
169. e all of them into a single Library ned that contains the NED declarations of all modules without all side effects However it is not recommended to put all object files into one library a or 1ib because then every simple module would be present in simulation programs linked with the library regardless whether the simulation uses them or not 5 7 2 Compound module NED source libraries The NED sources of reusable compound modules can also be placed in a library Candidates are e network nodes such as hubs bridges routers e different workstation computer types file server X terminal etc e node of a massively parallel multiprocessor used in testing different topologies e topology templates parameterized ring mesh hypercube torus etc topologies with the sizes shapes etc and the actual node types to be left as parameters The NED sources are used through the import mechanism the corresponding simple module object files still to be linked in the executable The user does not necessarily notice that he she is using a compound module library and not a simple module library In NED files the user imports and uses the compound module sources in exactly the same way as he she used the simple module interface declarations Linking also goes in the same way if the simple modules objects necessary for a certain compound module are aggregated into a library a or 1ib the user does not even notice the difference from the number of f
170. e grep for this purpose o grep queuelen queuingtime vector vec Or you can get the list of all vectors in the file by typing grep vector vector vec This will output the appropriate vector line vector 6 subnet 4 srvr queuelentqueuingtime 2 Pick the vector ID which is 6 in this case and grep the file for the vector s data lines grep 6 vector vec gt vector6 vec Now vector6 vec contains the appropriate vector The only potential problem is that the vector ID is there at the beginning of each line and this may be hard to explain to some programs that you use for post processing and or visualization This problem is eliminated by the OMNeT splitvec utility written in awk to be discussed in the next section 9 3 2 Using splitvec The splitvec script part of OMNeT breaks the vector file into several files which contain one vector each splitvec mysim vec creates several files mysiml vec mysim2 vec etc mysiml vec vector 1 subnet 4 term 12 response time 1 12 895 2355 66666666 14 126 4577 66664666 23 086 2355 66666666 mysim2 vec vector 2 subnet 4 srvr queuelen queuingtime 2 Technical University of Budapest Dept of Telecommunications OMNeT Manual Analyzing Simulation Results 174 16 960 2 00000000000 63663666 24 026 8 00000000000 44766536 As you can see the vector ID is gone The files can be further processed with math packages or read by analy
171. e implementation of models using Statistical Synchronization High level debugging is supported by saving the textual output from remote segments to a log file and or relaying them to a single console OMNeT supports flexible partitioning of the model In the NED language by using machine parameters you can specify logical hosts for different modules at any level of the module hierarchy of the network You map logical hosts to physical ones in the ini file if you map several logical hosts into the same physical machine they will be merged into a single OMNeT process The PVM3 Parallel Virtual Machine Version 3 library is used for communication between hosts PVM is portable and it is widely used in university and research environment You can find PVM readings in the Reference 10 1 2 Syncpoints Overview When running a simulation in parallel different segments of the model execute as independent UNIX processes typically on separate hosts Since the hosts can be of different speed and the simulated model segments can be of different complexity at a given moment the model times of different segments will differ some segments are ahead of the others and some lag behind Suppose that a message is sent from segment A to segment B which is ahead of A in model time If B processed the message causality would break This should never happen The solution built in OMNeT is the following Segment A must know in advance when it will send
172. e module to be included in the graph cPar yes yes topo extractByParameter include_in_topo amp yes The third form allows you to pass a function which can determine for each module whether it should or should not be included You can have cTopology pass supplemental data to the function through a void pointer An example which selects all top level modules and does not use the void pointer int select_function cModule mod void return mod gt parentModule simulation systemModule topo extractFromNetwork select_function NULL TBD one more example which does use the void ptr A cTopology object uses two types sTopoNode for nodes and sTopoLink for edges sTopoLinkIn and sTopoLinkOut are aliases forsTopoLink we ll speak about them later Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 114 Once you have the topology extracted you can start exploring it Consider the following code well explain it shortly for int i 0 i lt topo nodes i sTopoNode node topo node i ev lt lt Node i lt lt i lt lt is lt lt node gt module gt fullPath lt lt endl ev lt lt It has lt lt node gt outLinks lt lt conns to other nodes n ev lt lt and lt lt node gt inLinks lt lt conns from other nodes n ev lt lt Connections to other modules are n f
173. e processed in strict timestamp order in order to maintain causality that is to ensure that no event may have an effect on earlier events Processing an event involves calls to user supplied code For example using the computer network simulation example processing a timeout expired event may consist of re sending a copy of the network packet updating the retry count scheduling another timeout event and so on The user code may also remove events from the FES for example when cancelling timeouts Simulation stops when there are no more events left this happens rarely in practice or when it isn t necessary for the simulation to run further because the model time or the CPU time has reached a given limit or because the statistics have reached the desired accuracy At this time before the program exits the simulation programmer will typically want to record statistics into output files 5 1 3 Simple modules in OMNeT The user creates simple module types are by subclassing the cSimpleModule class which is part of the OMNeT class library cSimpleModule just as cCcompoundModulLe is derived from a common base class cModule cSimpleModule although stuffed with simulation related functionality doesnt do anything useful by itself The simulation programmer has to redefine some virtual member functions to make it do useful work These member functions are the following e void initialize e void activity e void han
174. e segmentation of the model for execution of different modules the compound module definition was extended with the machines and the on keywords Example module SomeNameForCompoundModule machines hostl1 host2 host3 host4 parameters gates submodules submodulel submodtypel on hostl submodule2 submodtype2 on host2 host3 submodule3 submodtypel on host4 connections endmodule The machines section lists formal host names which are used in the on lists of the submodules In the example the second submodule is itself a compound module that can be further subdivided to run on two separate hosts so its definition must have a machines section with two parameters Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 52 You do not have to propagate host names down to simple module level you can stop at a compound module which executes on a single host In other words a compound module with no machines section is equivalent to one with one machine parameter Of course you can give the same value to several machine parameters as to submodule1 s in the following example In this case the whole compound module will be placed on a single host as if it never had machine parameters at all module AnotherCompoundModule machines hostl host2 parameters gates submodules submodulel submodtypel on hostl hostl hostl Lit aes
175. e subclassing and e using shared objects 6 18 1 Using shared objects In a complex simulation a lot of messages are created sent and destroyed Messages typically have some parameters attached to them as cPar objects and it frequently happens that a certain parameter has identical values in all messages for example source address in a frame is the same in all messages sent by one module Still separate parameter objects are created and destroyed with each message which is very costly One could save significant amount of CPU time and memory if a single object could serve as a parameter to all existing messages This can be achieved with proper ownership control See the following example void MyComputer activity cPar source_addr address of this node cPar dest_addrs 3 possible destinations source_addr setStringValue DECnet000728 dest_addrs 0 setStringValue cisco_F99030 dest_addrs 1l setStringValue DEC__ 28E6AD dest_addrs 2 setStringValue DECnet000B04 Ne Ne Ne long k 0 for cMessage packet new cMessage DATA packet gt addPar new cPar sequence L k packet gt parList takeOwnership false NOTE THIS LINE packet gt addPar source_addr packet gt addPar dest_addrs k 3 send packet output gate wait truncnormal 1 5 0 5 Technical University of Budapest Dept of Telecommu
176. e the following awk program filter t DT print Do not forget the print statement or your filter will not output anything and the gnuplot graph will be empty Filters are automatically saved into and loaded from the ploverc file TBD add example scripts 9 2 Format of output vector files An output vector file contains several series of data produced during simulation The file is textual it looks like this mysim vec vector 1 subnet 4 term 12 response time 1 1 12 895 2355 66666666 1 14 126 4577 66664666 vector 2 subnet 4 srvr queuelentqueuingtime 2 2 16 960 2 00000000000 63663666 Technical University of Budapest Dept of Telecommunications OMNeT Manual Analyzing Simulation Results 173 1 23 086 2355 66666666 2 24 026 8 00000000000 44766536 There are label lines beginning with vector and data lines A vector line introduces a new vector Its columns are vector ID module of creation name of cOutVector object multiplicity of data single numbers or pairs will be written Lines beginning with numbers are data lines The columns vector ID current simulation time and one or two double values 9 3 Working without Plove 9 3 1 Extracting vectors from the file You can use the Unix grep tool to extract a particular vector from the file As the first step you must find out the ID of the vector You can find the appropriate vector line with a text editor or you can us
177. e valuel double value2 Records two values to the file It can be used only in the case if the instance of cOut Vector was created with tupl 2 otherwise it gives an error message void enableWrite Makes writing to the file enabled It is enabled by default void disableWrite Makes writing to the file disabled It is enabled by default void setStartTime simtime_t t Sets StartTime to t Data is recorded to the file only if simulation time gt StartTime The default value of StartTime is 0 void setStopTime simtime_t t Sets StopTime to t Data is recorded to the file only if the simulation time lt StopTime The default value of StopTime is 0 which means no Stoptime 15 11 2 class cOutFileMgr This class is responsible for handling the output file for cOutVectors Users don t need to use cOutFileMgr directly only through cOutVector Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 245 Constructor destructor copying cOutFileMgr const char s NULL Constructor virtual cOutFileMgr Destructor Closes the output file if it is still open Redefined virtual functions virtual const char className j Returns a pointer to the class name COutFileMgr Member functions void setFileName const char s Sets the name of the statistical output file This name will be set at the time of the next call of openFile void openFile
178. eT This chapter will be useful to those who want to extend the capabilities of the simulator or want to embed it into a larger application e The first two Appendices contain a comparison of OMNeT and two other important and well known simulation tools OPNET and PARSEC e Further Appendices provide reference of the NED language and the simulation library 1 4 History TBD 1 5 Authors OMNeT has been developed mostly by Andras Varga at the Technical University of Budapest Department of Telecommunications BME HIT Here s a list of the people who have or had anything to do with OMNeT Author and maintainer of the code Andras Varga BME HIT andras metechnology com andras whale hit bme hu Advisor contributor Gy rgy Pongor BME HIT pongor hit bme hu Co author old NED compiler until 1993 Akos Kun BME JAR compiler now called NEDC sample simulations summer 1995 Jan Heijmans TU Delft Alex Paalvast TU Delft Robert van der Leij TU Delft New feaures testing new examples fall 1995 Maurits Andr TU Delft M J A Andre twi tudelft nl George van Montfort TU Delft G P R vanMontfort twi tudelft nl Gerard van de Weerd TU Delft G vandeweerd twi tudelft nl JAR NEDC support for distributed execution also current user Gabor Lencse BME HIT lencse hit bme hu PVM support as final project spring 1996 Zoltan Vass BME HIT P k split algorithms and more from fall 1996 Babak Fakhamzadeh TU Delft Tec
179. ed into the put aside queue Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 224 Stopping the module or the simulation void end Ends the run of the simple module The simulation is not stopped unless this is the last running module void endSimulation Causes the whole simulation to stop Help for tracing and debugging bool snapshot cObject obj amp Simulation const char label NULL To be called from module functions Outputs textual information about all objects of the simulation including the objects created in module functions by the user into the snapshot file The output is detailed enough to be used for debugging the simulation by regularly calling snapshot one can trace how the values of variables objects changed over the simulation The arguments label is a string that will appear in the output file obj is the object whose inside is of interest By default the whole simulation all modules etc will be written out Tkenv also supports making snapshots manually from menu See also class cWatch and the WATCH macro void setPhase const char s Sets the phase string the function creates a copy of the string The string can be displayed in user interfaces which support tracing debugging currently only Tkenv and the string can contain information that tells the user what the module is currently doing const char phas
180. edefined byte pattern Oxdeadbeef at the stack boundary and reports stack violation if it was overwritten The mechanism usually works fine but occasionally it can be fooled by large and not fully used local variables e g char buffer 256 if the byte pattern happens to fall in the middle of such a local variable it may be preserved intact and OMNeT does not detect the stack violation To be able to make a good guess about stack size you can use the stackUsage call which tells you how much stack the module actually uses It is most conveniently called from finish void FooModule finish ev lt lt stackUsage lt lt bytes of stack used n The value includes the extra stack added by the user interface library see extraStackforEnvir in envir omnetapp h which is currently 8K for Cmdenv and 16K for Tkenv stackUsage also works by checking the existence of predefined byte patterns in the stack area so it is also subject to the above effect with local variables 6 17 Changing the network graphics at run time Sometimes it is useful to change the appearance of some components in the network graphics such as the color of the modules color width of connection arrows etc The appearance of nodes and connections is determined by the display strings Display strings are initially taken from the NED description stuff like display p 100 10 i pc You can change the display string of a module o
181. editors to define the network however this is only a good solution if there is an alternative form of topology description e g text file which allows one to generate the topology by program OPNET follows a unique way the network topology is stored in a proprietary binary file format which can be generated and read by the graphical editor and C programs linked against a special library On the other hand most non commercial simulation tools do not provide explicit support for topology description one must program a driver entity which will boot the model by creating the necessary nodes and interconnecting them Parsec SMURPH NS Finally a large part of the tools that do support explicit topology description supports only flat topologies CLASS OMNeT probably uses the most flexible method it has a human readable textual topology description format the NED language which is easy to create with any text processing tool perl awk etc and the same format is used by the graphical editor It is also possible to create a driver entity to build a network at run time by program OMNeT also supports submodule nesting Programming Model What is the programming model supported by the simulation environment Network simulators typically use either thread coroutine based programming such as activity in OMNeT or FSMs built upon a handleMessage like function For example OPNET SMURPH and NetSim use FSMs with underlying handleMes
182. eduler CScanScheduler for c in all_cache_managers do for s in all_schedulers do echo Parameters echo filesystem generator_type GenerateFromFile echo filesystem iolibrary_type PassThroughIOLibrary echo filesystem syscalliface_type PassThroughSysCalliIface echo filesystem filesystem_type PassThroughFileSystem echo filesystem cache_type c echo filesystem blocktranslator_type NoTranslation echo filesystem diskscheduler_type s echo filesystem accessmanager_type MutexAccessManager echo filesystem physicaldisk_type HP97560Disk gt algorithms ini filesystem done done And omnetpp ini includes algorithms ini Variations over seed value multiple independent runs The same technique can be used if you want several runs with different random seeds This code should do 500 runs with independent seeds suppose one run doesn t use more than 10 million random values bin bash for seed in seedtool g 1 10000000 500 do echo General echo random seed S seed echo output vector file xcube S seed vec gt parameters ini xcube done omnetpp ini should include parameters ini Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 165 Other languages for writing the control script The above examples use the Unix shell but you have quite
183. ented in OMNeT is very special and slightly odd The passed function is called for each object twice once on entering and once on leaving the object In addition after the first entering call to the function it signals with its return value whether it wants to go deeper in the contained objects or not Functions passed to forEach will use static variables to store other necessary information Yes this limits their recursive use forEach takes a function do_fn of ForeachFunc type with 2 arguments a cObject and a bool First forEach should call do_fn this true to inform the function about entering the object Then if this call returned true it must call forEach do_fn for every contained object Finally it must call do_fn this false to let do_fn know that there s no more contained object Functions using forEach work in the following way they call do_fn NULL false lt additional args gt to initialize the static variables inside the function with the additional args passed Then they call forEach do_fn for the given object Finally read the results by calling do_fn NULL false lt additional args gt where additional args can be pointers where the result should be stored ForeachFuncs mustn t call themselves recursively Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 218 I know this all is kind of weird but I wrote i
184. entries are Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 161 General ini warnings yes Helps debugging of the ini file If turned on OMNeT prints out the name of the entries it that it wanted to read but they were not in the ini file network The name of the network to be simulated distributed no Parallel execution or not number of runs 1 OMNeT will run multiple runs only of you set this value greater then 1 the existence of sections Run 1 Run 2 etc alone is not enough snapshot file omnetpp sna Name of the snapshot file The result of each snapshot call will be appended to this file output vector file omnetpp vec Name of output vector file output scalar file omnetpp sca Name of output scalar file pause in sendmsg no Only makes sense with step by step execution If enabled OMNeT will split send calls to two steps warnings yes Globally turns on off simulation runtime warnings It is advisable to leave this turned on log parchanges no Specifies whether changes of module parameters should be logged to file parchange file omnetpp pch File to save parameter changes to sim time limit 1000ms Duration of the simulation in simulation time cpu time limit 180s Duration of the simulation in real time random seed 542 Random number seed for generator 0 Should not be Zero total stack kb 8192 Specifies
185. ents of the simulator are executed by separate processors they have their own independent virtual times Because the interactions among segments are performed by the statistical parameters of these interactions the segmentation should be done so that the overwhelming majority of the interactions should happen within the segments and not among them This speeds up the so called inter segment transients and improves the accuracy as well Timing of statistics exchange Asynchronous statistics exchange means that whenever a statistical result collection in an output interface is ready it is applied after mapping and correction in the proper input interface This is clearly more efficient than the so called synchronous statistics exchange which means that we delay the application of collected values until all the output interfaces get ready with the result collection Frequent statistics exchange makes the inter segment transient faster but the lower sample numbers makes the estimation and the whole simulation less precise To learn more about SSM see PON92 and PON93 10 2 2 Using SSM in OMNeT OMNeT directly supports the implementation of statistical interface with the following classes cLongHistogram cDoubleHistogram cPSquare cPar Technical University of Budapest Dept of Telecommunications OMNeT Manual The Design of OMNeT 180 11 THE DESIGN OF OMNET 11 1 Structure of an OMNeT executable Consid
186. er interface that compiles and runs on all platforms whether it is UNIX DOS or WinNT console Cmdenv is designed primarily for batch execution Cmdenv uses simply executes all simulation runs that are described in the configuration file If one run stops with an error message subsequent ones will still be executed Cmdenv recognizes the following ini file options Cmdenv runs to execute 1 4 6 8 module messages no verbose simulation no display update 100ms The first one specifies which runs described in the Run 1 Run 2 etc sections should be executed If the value is missing Cmdenv executes all runs that have ini file sections if no runs are Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 166 specified in the ini file Cmdenv does one run The r command line option overrides this ini file setting The second and the third are yes no settings and control the amount of screen output during simulation The fourth one is in effect when the other two are disabled that is there would be no display at all from the simulation execution it prints out progress messages at the specified frequency Portability all platforms 8 5 2 Tkenv graphical user interface on Unix NT Features Tkenv is a portable graphical windowing user interface Tkenv supports interactive execution of the simulation tracing and debugging Tkenv is recommended in the developme
187. er of the entity Parsec s message buffer is practically the same as the put aside queue in OMNeT Guards may contain the special expressions qhead msgtype qempty msgtype qlength msgtype which refer to the messages in the message buffer The programmer perceives as if each message type had a separate message buffer receiv Request req when qhead Request units lt units receive Request req when qempty Release amp amp req units lt units Note that the qhead qempty and qlength operations seem to be all you can do with the message buffer while in OMNeT you have free access to the put aside queue through the cQueue member functions Parsec also has a hold statement which is functionally equivalent to OMNeT swait hold 5 Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix B Parsec and OMNeT 202 Cancelling messages Parsec has no support for cancelling messages that is there is no equivalent to OMNeT s cancelEvent The Parsec team recommends different workarounds like keeping a list of valid timers and checking messages against that A posting from the Parsec bulletin board Re Cancelling messages in PARSEC Posted by Rich from al bundy cs ucla edu on November 30 1999 Chano Gomez posted gt I would like to be able to cancel a message that has been already sent gt but that hasn t been received by the recipient because the messag
188. er the following diagram Executing SIM Model Model Component Library Fig 11 1 Architecture of OMNeT simulation programs A simulation program contains the simulated network with its simple and compound modules etc SIM ENVIR and exactly one of CMDENV and TKENV SIM contains the simulation class library and the simulation kernel The model only interacts with SIM ENVIR contains code that s common for all user interfaces and provides infrastructure like ini file handling for them main is also in ENVIR Specific user interface code is contained in CMDENV and TKENV The above components are also physically separated they are in separate source directories and form separate library files libsim_std a libenvir a etc The simulation program may contain several linked in model components networks simple module types compound module types channel types etc Any network but only one at a time can be set up for simulation which has all necessary components linked in 11 2 Embedding OMNeT Embedding is a special issue You probably do not want to keep the appearance of the simulation program so you do not want Cmdenv and Tkenv You may or may not want to keep ENVIR What you ll absolutely need for a simulation to run is the SIM package You can keep ENVIR if its philosophy and the infrastructure it provides omnetpp ini certain command line options etc fit into your desig
189. ere are two classes cOutVector and cOutFileMgr cOutFileMgr and cOutVector work together to allow the user save several series or output vectors a series of numbers or number pairs which are produced during simulation as results to a common file called output vector file or simply output file Users don t need to use cCOut FileMgr directly There is only one instance of cOutFileMgr in cSimulation it stores the common file pointer and gives out unique identifiers to cOut Vector objects There can be several cOut Vectors each one handles one output vector Data are written into a common file The output file is ASCII and can be read by a spreadsheet etc It should cause no problem to sort out values for one particular output vector 15 11 1 class cOutVector cOutVector is responsible for writing simulation data an output vector to a file A cOutVector object can write doubles or pair of doubles to the statistical output file The file consists of label lines and data lines Here are some sample lines vector 5 subnet 4 term 12 response time 1 5 12 895 2355 66666666 5 14 126 4577 66664666 vector 6 subnet 4 srvr queuelentqueuingtime 2 6 16 960 2 00000000000 63663666 5 23 086 2355 66666666 6 24 026 8 00000000000 44766536 There are label lines beginning with vector and data lines A label line introduces a new vector The columns vector vector ID module of creation n
190. es output vector files output scalar files and possibly the user s own output files OMNeT provides a GUI tool named Plove to view and plot the contents of output vector files But it is not expected that someone will process the result files using OMNeT alone output files are text files in a format which maybe after some preprocessing using sed awk or perl can be read into math packages like Matlab or its free equivalent Octave or imported into spreadsheets like Excel All these external programs have rich functionality for statistical analysis and visualization and OMNeT does not try to duplicate their efforts This manual briefly describes some data plotting programs and how to use them with OMNeT User interfaces The primary purpose of user interfaces is to make the inside of the model visible to the user to start stop simulation execution and possibly allow the user intervene by changing variables objects inside the model This is very important in the development debugging setPhase of the simulation project Just as important a hands on experience allows the user to get a feel about the model s behaviour A nice graphical user interface can also be used to demonstrate how the model works internally The same simulation model can be executed with different user interfaces without any change in the model files themselves The user would test and debug the simulation with a powerful graphical user interface and finally ru
191. ess will start up the program on the other hosts too and do the simulation The first machine is called console or master the others are called slaves 10 1 7 If there are problems PVM programs in general are more difficult to get running than ordinary programs Wrong settings in the PVM configuration files can cause various problems for example Also parallel programs are a lot harder to test and debug What can you do if your distributed OMNeT simulation won t work e First of all check the pvmhosts file to see if PVM looks for the executables in the right directories on all hosts and the working directories are right typically the same directory as the executable s e In the ini file enable writing the slave log files for the slave processes and check what is written into them e You can try enabling the SINGLE_HOST define in the sim pvm pvmmod cc source file This will make OMNeT run all segments of the distributed simulation on the local host making things a lot easier to manage e Also try the defining PVM_DEBUG at the same place it enables a lot of ev printf s in the code interfacing with PVM so it is easier to spot where the problems are e PVM itself also has an environment variable which if set causes the PVM library to print out debugging information However this is very low level information it will rarely be useful 10 2 Statistical synchronization 10 2 1 The description
192. ete you can set a breakpoint to that function If COUNTBLOCKS is turned on you should see the heap cc DEBUG ALL BLOCKS FREED OK message at the end of the simulation If you do not see it it means that some blocks have not been freed up properly that is your simulation program is likely to have memory leaks 8 7 Execution speed If your simulation program is tested and runs OK youll probably want to run it as fast as possible Here s a table that could help where to begin optimizing Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 170 The measurements were made on one version of the FDDI model you can find it in the samples directory we simulated 10 milliseconds We used Cmdenv The machine was a 100Mhz Intel Pentium with 32MB RAM The simulation program was compiled with Borland C 3 1 no particular optimization and run on DOS 6 22 Disk caching was installed SmartDrive read write caching 8MB cache Settings Execution Details time all screen output on 7 min 50 sec Setting in omnetpp ini full heapcheck verbose simulation yes module messages yes The defines in envir cmdenv heap cc were all enabled This means full heapcheck with each allocation tracking of all allocated blocks etc no screen output at 5 min 50 sec All screen output were ifdef ed out from source all also the omnetpp ini contained the full heapcheck
193. eter to the compound module module PC parameters mac_type string submodules mac mac_type like General_MAC OK endmodule The mac_type parameter should take the value TokenRing_MAC Ethernet_MAC or FDDI_MAC and a submodule of the appropriate type will be created The value for the parameter can even be given in the ini file This gives you a powerful tool to customize simulation models see also Topology templates from the NED chapter 5 2 4 The class declaration As mentioned before simple module classes have to be derived from cSimpleModule either directly or indirectly In addition to overwriting some of the previously mentioned four member functions initialize activity handleMessage finish you have to write a constructor and some more functions Some of this task can be automated so when writing the C class declaration you have two choices 1 either use a macro which expands to the stock version of the functions 2 or write them yourself Using macro to declare the constructor If you choose the first solution you use theModule_Class_Members macro Module_Class_Members classname baseclass stacksize The first two arguments are obvious baseclass is usually cSimpleModule but stacksize needs some explanation If you use activity the module code runs as a coroutine so it will need a separate stack This will be discussed in detail later Technical University of Buda
194. eters can be included with the name notation p xpos Sypos b rect 60 10 0 Sfillcolor black 2 Objects that may have display strings are e compound modules as the enclosing module in the drawing e submodules e connections Tags used in submodule display strings Tag Meaning P xpos ypos Place submodule at xpos ypos pixel position with the origin being the top left corner of the enclosing module Defaults an appropriate automatic layout is where submodules do not overlap If applied to a submodule vector ring or row layout is selected automatically p xpos ypos row deltax Used for module vectors Arranges submodules in a row Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 57 p xpos ypos col deltay p xpos ypos matrix itemsperrow deltax deltay Pp xpos ypos ring width height b width height rect b width height oval o fillcolor outlinecolor borderwidth i iconname Examples p 100 60 i workstation p 100 60 b 30 30 rect 0 4 starting at xpos ypos keeping deltax distances Defaults deltax is chosen so that submodules do not overlap Used for module vectors Arranges submodules in a column starting at xpos ypos keeping deltay distances Defaults deltay is chosen so that submodules do not overlap Used for module vectors Arranges submodules in a matrix starting at x
195. f the art commercial simulation program for the modeling of communication systems OPNET is designed to enable full detail modeling every tool is given to implement nonstandard protocols or behaviour A quote from the OPNET brochure e OPNET presents an advanced graphical user interface that supports multi windowing makes use of menus and icons and runs under X Windows Supported platforms include popular engineering workstations from SUN DEC HP and Silicon Graphics Windows NT version also exists Graphical object oriented editors for defining topologies and architectures directly parallel actual systems allowing an intuitive mapping between a system and its model OPNET s hierarchical approach simplifies the specification and representation of large and complex systems e The process editor provides a powerful and flexible language to design models of protocols resources applications algorithms queuing policies and other processes Specification is performed in the Proto C language which combines a graphical state transition diagram approach with a library of more than 300 communication and simulation specific functions The full generality and power of the C language is also available OPNET simulations generate user selected performance and behavioral data Simulation results can be plotted as time series graphs scatter plots histograms and probability functions Standard statistics and confidence intervals are easily ge
196. ffic modeling e multi processor and distributed systems e administrative systems e and any other system where the discrete event approach is suitable An OMNeT model consists of hierarchically nested modules The depth of module nesting is not limited which allows the user to reflect the logical structure of the actual system in the model structure Modules communicate with message passing Messages can contain arbitrarily complex data structures Modules can send messages either directly to their destination or along a predefined path through gates and connections Modules can have parameters which are used for three main purposes to customize module behaviour to create flexible model topologies where parameters can specify the number of modules connection structure etc and for module communication as shared variables Modules at the lowest level of the module hierarchy are to be provided by the user and they contain the algorithms in the model During simulation execution simple modules appear to run in parallel since they are implemented as coroutines sometimes termed lightweight processes To write simple modules the user does not need to learn a new programming language but he she is assumed to have some knowledge of C programming OMNeT simulations can feature different user interfaces for different purposes debugging demonstration and batch execution Advanced user interfaces make the inside of the model visible t
197. file ssseseeeseeeeseeeeseeressesssesressessresreesersreseresreseese 157 8 2 3 Module parameters in the configuration file sseeeseeeseseeeeeseeeseerresersrrsrresreseeee 158 8 2 4 Configuring Output VEClOLS 43 Bivatiy ie ines eevee eel ea eee eae Sse 158 8 2 5 Module parameter logging aii ciennrt ties i eendes eee a tore etnias 159 89 2 0 Display Sting S rsa ed ge aya gle de TEA ssp aie OL eae Ahh 159 Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 9 8 2 1 SPECILyINE seed vales artnr i suey eaatarye e a save E dems Ea 160 9 2 9 Listof all ini fle options eia a a a as 160 8 3 Choosing good seed values the seedtool utility ssssssesessesessesseessesssesesseeesseesseesseesseee 163 8 4 Repeating or iterating simulation runs s sesessseessessseesseressettsseessesseesseeesseeesseesseesseesseee 163 8 5 User interfaces of simulation executables cee ceesccceeceeceseeeceeeeeceeeeecseeeeeeeeeeseeeenaeeeenae 165 8 5 1 Cmdenv the command line user interface eeecceceeceeeeeeeeceteeeeseeeenseeeenes 165 8 5 2 Tkenv graphical user interface On Unix NT 0 ee ceecceeeeeeeeeeeeeeeeeeeeeneeeeneeeenes 166 8 52 Ee VISTA AM et nnsa uescand Soaaspds N E a 168 8 6 Typical Problem Sisti ar e e aa ete Gs dbed aaa aao sta e a 168 8 6 1 Stack problems nsii a A a aE Oa A a EA 168 8 6 2 Memory allocation problems sssssseessseeeeseeessessesssereseeessseessesseessee
198. former ones Thus values intended as defaults should appear in the first sections Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 43 4 5 4 Connections In a compound module definition the gates of the compound module and its immediate submodules are connected In other words the NED language does not support connections that would cross the walls of a compound module without using gates of that module Only point to point connections are supported In summary 1 The gate of a submodule or enclosing module gate can be connected another submodule or enclosing module gate 2 Gate direction must be observed e g you cannot connect two submodule output gates Connections are specified in the Connections section of a compound module definition It lists the connections separated with semicolons Example module SomeCompound parameters gates submodules connections nodel output gt node2 input nodel input lt node2 output TD So endmodule Each connection can be e direct that is no delay bit error rate or data rate can use a named channel or a channel given with delay error and data rate values e single or multiple loop connection e conditional or non conditional These connection types are described in the following sections Single connections and channels The source gate can be an output gate of a submodule or an inp
199. function returns the number of items in the queue and empty tells whether there s anything in the queue There are other functions dealing with insertion and removal The insertBefore and insertAfter functions insert a new item exactly before and after a specified one regardless of the ordering function The tail and head functions return pointers to the objects at the tail and head of the queue without affecting queue contents The pop function can be used to remove items from the tail of the queue and the remove function can be used to remove any item known by its pointer from the queue queue remove msg Priority queue By default cQueue implements a FIFO but it can also act as a priority queue that is it can keep the inserted objects ordered If you want to use this feature you have to provide a function that takes two cOb ject pointers compares the two objects and returns 1 0 or 1 as the result see the reference for details An example of setting up an ordered cQueue cQueue sortedqueue Sortedqueue cObject cmpbyname true sorted by object name ascending If the queue object is set up as an ordered queue the insert function uses the ordering function it searches the queue contents from the head until it reaches the position where the new item needs to be inserted and inserts it there Iterators Normally you can only access the objects at the head or tail of the queue However
200. gate int routeContains cModule m int g 1 Determines if a given module is in the route that this gate is in bool isRouteOK Returns t rue if the route that this gate is in is complete i e if it starts and arrives at a simple module void deliver cMessage msg This function is called internally by the send functions to deliver the message to its destination 15 8 Value holding class cPar Header file coar h sim directory Parameter classes are designed to hold a value Many types are available e string e boolean e long int e double e function e expression e distribution e void pointer e object pointer e indirect value For all types an input flag can be set In this case the user will be asked to enter the value when the object s value is first used The prompt string can also be specified for cPar If no prompt string is given the object s name will be displayed as prompt text Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 230 15 8 1 class cPar Construction destruction copying cPar cPar amp other Copy constructor creates an exact copy of the argument cPar const char name NULL Constructor creates a cPar with the given name and long L as default type cPar const char name cPar amp other Constructor creates a copy of the second argument with another name virtual cObject dup
201. gh the gate given with its name and index if multiple gate Parameters and attached objects cArray parList Returns the cArray member of the message which holds the parameters and other attached objects Parameters can be inserted retrieved looked up or deleted through cArray s member functions cPar amp addPar cPar amp p cPar amp addPar const char name Convenience functions add a parameter to the message s parameter list cPar amp par int index Convenience function returns the indexth object in the message s parameter list converting it toacPar cPar amp par const char name Convenience function returns the object with the given name in the message s parameter list converting it toa cPar int hasPar const char name Convenience function returns true if there s an object with the given name in the message s parameter list int findPar const char name Convenience function returns the index of the parameter with the given name in the message s parameter list or 1 if it could not be found Miscellaneous functions static int cmpbydelivtime cObject one cObject other Static function that compares two messages by their delivery times then by their priorities static int cmpbypriority cObject one cObject other Static function that compares two messages by their priority It can be used to sort messages in a priority queue 15 6 2 Network packet class cPacket cPacke
202. go n ev lt lt There are lt lt node gt paths lt lt equally good directions taking the first one n sTopoLinkOut path node gt path 0 ev lt lt Taking gate lt lt path gt localGate gt fullName lt lt we arrive in lt lt path gt remoteNode gt module gt fullPath lt lt on its gate lt lt path gt remoteGate gt fullName lt lt endl node path gt remoteNode The purpose of the distanceToTarget member function of a node is self explanatory In the unweighted case it returns the number of hops The paths member function returns the number of edges which are part of a shortest path and path i returns the ith edge of them as sTopoLinkOut If the shortest paths were created by the SingleShortestPaths function paths will always return 1 or 0 if the target is not reachable that is only one of the several possible shortest paths are found The MultiShortestPathsTo functions find all paths at increased run time cost The cTopology s targetNode function returns the target node of the last shortest path search You can enable disable nodes or edges in the graph This is done by calling their enable or disable member functions Disabled nodes or edges are ignored by the shortest paths calculation algorithm The enabled member function returns the state of a node or edge in the topology graph One usage of disable is when you want
203. grams Technical University of Budapest Dept of Telecommunications OMNeT Manual Building Simulation Programs 148 simulation kernel library user network impl of simple interface libraries lib a description ned INEDC compiling impl of modules cc lib a structure n cc C compiling po simulation program configuration file omnetpp ini execution output files vec sna etc This section discusses how to use the simulation system on the following platforms e Unix with gcc installed and which is similar Cygwin on Windows NT e Borland C 5 0 on Windows NT e MSVC 6 0 on Windows NT 7 2 Using Unix and gcc 7 2 1 Installation Installation essentially consists of building NEDC and the necessary libraries After you ve unpacked the distribution in your home directory just type the following commands cd omnetpp configure make It will build all the libraries and the example programs for you If you have a C compiler other then gcc you ll have to edit theconfigure script Technical University of Budapest Dept of Telecommunications OMNeT Manual Building Simulation Programs 149 7 2 2 Producing a makefile with the makemake script The makemake script can automatically generate the makefile for your simulation program based on the source files it finds in your directory makemake has several options type makemake h to see them
204. guage Grammar expression expression AN expression expression expression expression expression expression expression lt expression expression lt expression expression gt expression expression gt expression expression expression expression expression AND expression expression OR expression NOT expression expression functionname expression expression numconstvalue inputvalue ANCESTOR REF parametername SIZEOF gatename INDEX numconstvalue integerconstant realconstant timeconstant otherconstvalue characterconstant stringconstant TRUE FALSE inputvalue INPUT default prompt string default expression otherconstvalue used with distributed execution used with the statistical synchronization method ook Define_Function macro Size of a vector gate ok ok 2 kK x Index in submodule vector aokk a ok Can appear in any order Technical University of Budapest Dept of Telecommunications 209 max three arguments The function name must be declared in the C sources with the OMNeT Manual Appendix D Class Library Reference 210 15 APPENDIX D CLASS LIBRARY REFERENCE 15 1 What is documented here This is a fairly complete reference of the object library of OMNeT Many details are not covered in this documentation Only public member fun
205. h gt versions char opp_strdup const char Duplicates the string If the pointer passed is NULL or points to a null string NULL is returned char opp_strcpy char const char Same as the standard st rcpy function except that NULL pointers in the second argument are treated like pointers to a null string int opp_strcmp const char const char Same as the standard strcmp function except that NULL pointers are treated like pointers to a null string bool opp_strmatch const char const char Returns t rue if the two strings are identical up to the length of the shorter one NULL pointers are treated like pointers to a null string const char correct const char Returns the pointer passed as argument unchanged or if it was NULL returns a pointer to a null string char opp_concat const char sl const char s2 const char s3 NULL const char s4 NULL Concatenates 2 3 or 4 strings and places the result into a static buffer and returns the buffer s pointer The result s length shouldn t exceed 255 characters char indexedname char buf const char name int index Creates a string like component 35 into buf the first argument 15 2 5 Basic random number generation OMNeT has a built in pseudorandom number generator that gives long int 32 bit values Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library
206. h simply stores the new state in the cFSM object FSM_Goto fsm newState The FSM starts from the state with the numeric code 0 this state is conventionally named INIT Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 82 Debugging FSMs If you define FSM_DEBUG before including omnetpp h each state transition will be logged to ev define FSM_DEBUG include lt omnetpp h gt The actual printing is done through the FSM_Print macro You might redefine it if you dont like what it currently does define FSM_Print fsm exiting ev lt lt FSM lt lt fsm name lt lt exiting exiting entering lt lt fsm stateName lt lt endl Implementation The FSM_Switch is a macro It expands toa switch statement embedded in a for loop which repeats until the FSM reaches a steady state The actual code is rather ugly but if youre dying to see it it s incfsm h Infinite loops are avoided by counting state transitions if an FSM goes through 64 transitions without reaching a steady state the simulation will terminate with an error message An example Let us write another flavour of a bursty generator It has two states SLEEP and ACTIVE In the SLEEP state the module does nothing In the ACTIVE state it sends messages with a given inter arrival time The code was taken from the fi fo2 sample simulation defi
207. hat remote and local can be correctly interpreted for edges of both directions 6 8 3 Shortest paths The real power of cTopology is in finding shortest paths in the network to support optimal routing cTopology finds shortest paths from all nodes to a target node The algorithm is computationally inexpensive In the simplest case all edges are assumed to have the same weight A real life example when we have the target module pointer finding the shortest path looks like this sTopoNode targetnode topo nodeFor targetmodule topo unweightedSingleShortestPathsTo targetnode Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 115 This performs the Dijkstra algorithm and stores the result in the cTopology object The result can then be extracted using cTopology and sTopoNode methods Naturally each call to unweightedSingleShortestPathsTo overwrites the results of the previous call Walking along the path from our module to the target node sTopoNode node topo nodeFor this if node NULL ev lt lt We lt lt fullPath lt lt are not included in the topology n on if node gt paths 0 ev lt lt No path to destination n else while node topo targetNode ev lt lt We are in lt lt node gt module gt fullPath lt lt endl ev lt lt node gt distanceToTarget lt lt hops to
208. he lower or upper limit and have the other end of the range estimated automatically The setRange member functions of cDensityEstBase deal with setting up the histogram range It also provides pure virtual functions transform etc Subsequent observations are placed in the histogram structure If an observation falls out of the histogram range the underflow or the overflow cell is incremented Construction destruction copying cDensityEstBase cDensityEstBase r cDensityEstBase const char s NULL virtual cDensityEstBase cDensityEstBase amp operator cDensityEstBase amp res Constructors destructor assignment Redefined virtual functions virtual const char className virtual void writeContents ostream amp os virtual void collect double val virt virt virt ual void clearResult ual void saveToFile FILE ual void loadFromFile FILE pa i Os OL New virtual functions virtual bool transformed virtual void transform First one returns whether the object is transformed or not second one forces a transformation virtual unsigned cells virtual double basepoint unsigned k virtual unsigned cell unsigned k Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 250 virtual double cellPDF unsigned k
209. he message receiving functions can only be used in the activity function handleMessage gets the messages in its argument list A message can be received by a number of functions the most general one is the receive function cMessage msg receive Simple module objects contain a built in queue object called putAsideQueue The put aside queue is used by some of the message receiving functions There are two groups of functions that receive messages e receive receiveOn e receiveNew receiveNewOn The functions receive receiveOn check the put aside queue first and try to return a message from it Only if they do not find an appropriate message in the put aside queue will wait for new messages The functions receiveNew receiveNewOn wait for new messages ignoring the put aside queue Furthermore the On functions expect messages to arrive on a specific gate Messages that arrive on another gate are inserted the put aside queue The On less versions accept any message Since the receive and receiveOn return messages also from the put aside queue the arrival times of messages they return can be less than the current simulation time A naive and also incorrect approach to check whether a message is a new one or it has been retrieved from the putaside queue could be the following cMessage msg receive if msg gt arrivalTime lt simTime not correct several events may occ
210. he output vector is there Plove can plot it for you Moreover it can also work with gzipped vector files without extracting them just make sure you have zcat Plove never modifies the output vector files themselves On startup Plove automatically reads the ploverc file in your home directory The file contains general gnuplot settings the filter configuration etc that is the stuff from the Options menu Portability Plove works fine on Unix and with some limitations on Win95 NT 9 1 2 Usage First you load an output vector file vec into the left pane You can also load gzipped vector files vec gz without having to decompress them You can copy vectors from the left pane to the right pane by clicking the right arrow icon in the middle The large PLOT button will plot the selected vectors in the right pane Selection works as in Windows dragging and shift left click selects a range and ctrl left click selects deselects individual items To adjust drawing style change vector title or add filter push the Options button This works for several selected vectors too Plove accepts nc mc like keystrokes F3 F4 F5 F6 F8 grey and grey The left pane works as a general storage for vectors you re working with You can load several vector files delete vectors you dont want to deal with rename them etc All this will not affect the vector Technical University of Budapest Dept of Telecommunications OMNeT
211. he source for simulation kernel is not available although the ready made protocol models come with sources In conclusion it can be said that OMNeT has enough features to make it a good alternative to most network simulation tools and it has a strong potential to become one of the most widely used network simulation packages in academic and research environments The most serious shortcoming is the lack of available protocol models but since this is mostly due to the fact that it is a relatively new simulation tool with the help of the OMNeT user community the situation is likely to become much better in the future 1 3 Organization of this manual The manual is organized around the following topics e The chapters 1 2 and 3 contain introductory material some overview and an example simulation e The second group of chapters 4 5 and 6 are the programming guide They present the NED language the simulation concepts and their implementation in OMNeT explain how to write simple modules and describe the class library e The following chapters 7 8 and 9 deal with practical issues like building and running simulations and analyzing results and present the tools OMNeT has to support these tasks e Chapter 10 is devoted to the support for distributed execution Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 17 e Finally Chapter 11 explains the architecture and the internals of OMN
212. hical interface You load the appropriate file by selecting in a dialog box The icon bar and menu commands can be used to customise the graph Some say that Xmgr can produce nicer output that Gnuplot and it is easier to use Xmgr cannot do 3D and only runs on Unixes with X and Motif installed Xmgr also has a batch interface so you can use it from scripts too Technical University of Budapest Dept of Telecommunications OMNeT Manual Parallel Execution 175 10 PARALLEL EXECUTION 10 1 Support for parallel execution 10 1 1 Overview OMNeT supports parallel execution of large simulations To make use of parallel execution the model is to be partitioned to several segments that will be simulated independently on different hosts or processors The simulation kernel makes it possible to send messages from one segment to another A message can contain arbitrarily complex data structures these are transferred transparently even between hosts of different architectures The simulation kernel provides a simple synchronization mechanism syncpoints available through the syncpoint call that can ensure that causality is kept when sending messages between segments Syncpoints correspond to null messages found in the literature Message sending and syncpoints enable one to implement conservative PDES and also a novel and less known paradigm Statistical Synchronization The simulation class library contains objects that explicitly support th
213. hnical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 18 We have to mention Dr Leon Rothkranz from the Technical University of Delft whose work made it possible for the Delft students to come to Budapest in 1995 The starting point of this manual was the 1995 report of Jan Heijmans Alex Paalvast and Robert van der Leij Technical University of Budapest Dept of Telecommunications OMNeT Manual Overview 19 2 OVERVIEW 2 1 Modeling concepts OMNeT provides efficient tools for the user to describe the structure of the actual system Some of the main features are e hierarchically nested modules e modules are instances of module types e modules communicate with messages through channels e flexible module parameters e topology description language 2 1 1 Hierarchical modules An OMNeT model consists of hierarchically nested modules which communicate with messages OMNeT models are often referred to as networks The top level module is the system module The system module contains submodules which can also contain submodules themselves Fig 2 1 The depth of module nesting is not limited this allows the user to reflect the logical structure of the actual system in the model structure simple modules Fig 2 1 Simple and compound modules Modules that contain submodules are termed compound modules as opposed simple modules which are at the lowest level of the module hierarchy S
214. how easily the rightNodelndex i j function can be formulated gt Enumerate All Connections A third pattern is to list all connections within a loop for i 0 Nconnections 1 do node leftNodeIndex i out gt node rightNodeIndex i in endfor The pattern can be used if leftNodeIndex i and rightNodelIndex i mapping functions can be sufficiently formulated The Serial module is an example of this approach where the mapping functions are extremely simple leftNodeIndex i i and rightNodelIndex i i 1 The pattern can also be used to create a random subset of a full graph with a fixed number of connections In the case of irregular structures where none of the above patterns can be employed the user can resort to specifying constant submodule gate vector sizes and explicitly listing all connections like he she would do it in most existing simulators 4 6 4 Topology templates Overview Topology templates are nothing more than compound modules where one or more submodule types are left as parameters using the like phrase of the NED language You can write such modules which implement mesh hypercube butterfly perfect shuffle or other topologies and you can use them wherever needed in you simulations With topology templates you can reuse interconnection structure An example hypercube The concept is demonstrated on a network with hypercube interconnection When building an N dimension hypercube we
215. ia aiias 149 7 2 4 Static vs shared OMNeT system libraries eceesceeeseeesseeeeeeeeeesteeeenaees 150 7 3 Using Win32 with Borland C 5 0 seesssseessesssesssesssseessseessrssessseresseeesseessesseessseesssees 150 7 3 1 Compiling the libraries and nedc ssssosessseesseessesssessseeessseessessersseessseersseesseesse 150 T32 Settings upa project le r e a aea ar E ae a A AE E E E aa 151 7 3 3 Startins from Scrat hiarr a a a a a E E S 151 FA Using WMI With MSV G ioega e aee or eae Eo NESTU Ras 152 7 4 1 Prerequisite install Tcl Tk ssesseneseeeseeeseseesseesseesseesseeesseeessressessserssseessseesseesse 152 74 2 Installing OMN T F ie ionar a s s t e a a aS 152 7 4 3 Building the samples from the MSVC IDE seesseseeeeesesesssesesressesrresressrseresresss 152 7 4 4 Creating project files for your SimULatiONnS eee eee eeseceeeeeseeeeseeenaeeneeeeees 153 7 4 5 Reconfiguring the MDraries 55 05 2025 sues ccasensetted ears usages ues aesadaaductente peas eins yeas 154 TAGUSE PL OVE erni e a r e a gees baw Snes stad aota ae aa i SaS 154 8 RUNNING THE SIMULATION ssssssrssssssosssssssotesssssosesesssotessssneos sses sots ss sssr stess sss 155 8 1 Command ine sWitehESss sesine aa i ie Pak poe cogs aiaa 155 8 2 The configuration file omnetpp ini sesesssesssseessesssessseeesseessseesseesseesseeesseeesseessresseeeseee 155 8 2 1 Sections and eties eni a E et E E ni avid 155 8 2 2 Splitting up the configuration
216. ied you ll get an error message cPar p new cPar par int index array addAt 5 p Finding an object in the array int index array find p Getting a pointer to an object at a given index cPar p cPar array index You can also search the array or get a pointer to an object by the object s name int index array find par j Par p cPar array par You can remove an object from the array by calling remove with the object name the index position or the object pointer array remove par array remove index array remove p The remove function doesn t deallocate the object but it returns the object pointer If you also want to deallocate it you can write delete array remove index Iteration cArray has no iterator but it s easy to loop through all the indices with an integer variable The items member function returns the largest index plus one for int i 0 i lt array items i if array i is this position used Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 121 cObject obj array il ev lt lt obj gt name lt lt endl 6 11 Non object container classes There are two container classes to store non object items cLinkedList and cBag The first one parallels with cQueue the second one with cArray They can be useful if you have to deal with C structs or objects that are not
217. igned char s inline cEnvir operator lt lt cEnvir amp ev const unsigned char s inline cEnvir amp operator lt lt cEnvir amp ev unsigned char c inline cEnvir amp operator lt lt cEnvir amp ev signed char c inline cEnvir amp operator lt lt cEnvir ev short i inline cEnvir amp operator lt lt cEnvir ev unsigned short i inline cEnvir amp operator lt lt cEnvir amp ev int i inline cEnvir amp operator lt lt cEnvir ev unsigned int i inline cEnvir operator lt lt cEnvir ev long 1 inline cEnvir amp operator lt lt cEnvir amp ev unsigned long 1 inline cEnvir amp operator lt lt cEnvir ev float f inline cEnvir amp operator lt lt cEnvir ev double d inline cEnvir amp operator lt lt cEnvir ev long double d Output operators for all fundamental data types inline cEnvir amp operator cEnvir amp ev const signed char s inline cEnvir amp operator cEnvir amp ev const unsigned char s The operator can be used to replace ev prompt For example ev How many gt gt n m inline cEnvir amp operator gt gt cEnvir amp ev signed char s inline cEnvir amp operator gt gt cEnvir amp ev unsigned char s inline cEnvir amp operator gt gt cEnvir amp ev char c inline cEnvir amp operator gt gt cEnvir amp ev unsig
218. iles he she has to link in 5 7 3 Precompiled compound module libraries If you share a compound module with others you do not necessarily have to share the NED source and reveal the internals of the compound module You can turn the compound module into something that very much looks like a simple module Suppose you have the following compound module router compound ned module Router parameters processing_delay buffersize gates in input_ports out output_ports submodules routing RoutingModule parameters ALa ses gatesizes hil tees Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 89 datalink DataLink num_of_ports parameters retry_count 5 window_size 2 EE connections endmodule First you would compile this NED file with the NEDC compiler and the resulting C code with the C compiler Then you would aggregate this object file with the simple module object files into a single library a or 1ib Also you would write a separate NED file that declares the interface of the new simple module couter simple ned simple Router parameters processing_delay buffersize gates in input_ports out output_ports endsimple The method produced a precompiled compound module The resulting two files can be placed into a simple module library and can be used identically to ordinary simple modules Using precompiled compound m
219. iles reside 4 3 Channel definitions A channel definition specifies a connection type of given characteristics The channel name can be used later in the NED description to create connections with these parameters Example channel DialUpConnection delay normal 0 004 0 0018 error 0 00001 datarate 14400 endchannel Any of the delay error and datarate parameters are optional and they can appear in any order The values are NED expressions This means that they can be constants integer or real random values from various distributions etc 4 4 Simple module definitions Simple modules are the basic building blocks for other compound modules A simple module is defined by declaring its parameters and gates Example simple SomeNameForModule parameters ER gates CEES endsimple 4 4 1 Simple module parameters Parameters are variables that belong to a module Simple module parameters can be queried and used by simple module algorithms For example a parameter called num_of_messages can be used by a module called MsgSource to determine how many messages it has to generate Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 37 Parameters are declared by listing their names in the parameters section of a module description The parameter type can optionally be specified as numeric numeric const or simply const bool string or anytype Example simple MsgSour
220. imple modules contain the algorithms in the model The user implements the simple modules in C using the OMNeT simulation class library 2 1 2 Module types Both simple and compound modules are instances of module types While describing the model the user defines module types instances of these module types serve as components for more complex module types Finally the user creates the system module as an instance of a previously defined module type all modules of the network are instantiated as submodules and sub submodules of the system module When a module type is used as a building block there is no distinction whether it is a simple or a compound module This allows the user to split a simple module into several simple modules Technical University of Budapest Dept of Telecommunications OMNeT Manual Overview 20 embedded into a compound module or vica versa aggregate the functionality of a compound module into a single simple module without affecting existing users of the module type Module types can be stored in files separately from the place of their actual usage This means that the user can group existing module types and create component libraries This feature will be discussed later in chapter 8 2 1 3 Messages gates links Modules communicate by exchanging messages In an actual simulation messages can represent frames or packets in a computer network jobs or customers in a queuing network or other type
221. impleModule Module_Class_Members Generator cSimpleModule 0 note zero stack size virtual void initialize virtual void handleMessage cMessage msq Define _Module Generator void Generator initialize schedule first sending scheduleAt simTime new cMessage Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 77 void Generator handleMessage cMessage msg generate amp send packet cMessage pkt new cMessage send pkt out schedule next call scheduleAt simTime exponential 1 0 msg Example 2 Bursty traffic generator A bit more realistic example is to rewrite our Generator to create packet bursts each consisting of burst_length packets We add some data members to the class e burst_length will store the parameter that specifies how many packets a burst must contain e burst_ctr will count in how many packets are left to be sent in the current burst The code class BurstyGenerator public cSimpleModule Module_Class_Members Generator cSimpleModule 0 note the zero stack size int burst_length int burst_ctr virtual void initialize virtual void handleMessage cMessage msq Define_Module BurstyGenerator void BurstyGenerator initialize init parameters and state variables burst_length par burst_length
222. imulation is executed and can also contain settings of model parameters The ini file is a text file consisting of entries grouped into different sections The following sections can exist General Cmdenv Tkenv Parameters OutVectors Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation DisplayStrings Machines Slaves Run 1 Run 2 Run 3 and denote comments A sample ini file omnetpp ini General ini warnings false network token distributed no snapshot file token sna output vector file token vec log parchanges no parchange file token pch random seed 1 sim time limit 1000ms cpu time limit 180s total stack kb 2048 Cmdenv runs to execute 1 3 5 module messages yes verbose simulation no display update 100ms Parameters token num_stations 3 token num_messages 10000 Run 1 token wait_time 10ms Run 2 token wait_time 30ms Parameters that were set to input value in the NED file are searched for in the ini file 156 OMNeT can execute several simulation runs automatically one after another If multiple runs are selected option settings and parameter values can be given either individually for each run or together for all runs depending in which section the option or parameter appears This is summarized in the following table
223. ined cObject functions void collect double val virtual void clearResult virtual long samples j virtual double min virtual double max virtual double mean virtual double stddev virtual double variance virtual void saveToFile FILE virtual void loadFromFile FILE Redefined cStatistic functions virtual double random Redefined cStatistic function cStdDev s random number generator returns numbers of normal distribution with the current mean and standard deviation 15 12 3 class cWeightedStdDev cWeightedStdDev is designed to collect doubles and calculate weighted statistics of them Construction destruction copying cWeightedStdDev cWeightedStdDev r cWeightedStdDev const char s NULL virtual cWeightedStdDev virtual cObject dup cWeightedStdDev operator cWeightedStdDev amp res Constructors destructor duplication and assignment Redefined virtual functions virtual const char className j Returns a pointer to the class name string cWeightedStdDev virtual void info char buf virtual const char inspectorFactoryName virtual void writeContents ostream amp os Redefined cObject functions void collect2 double val double weight New member function vir ual void clearResult virtual long weights virtual double min virtual
224. int gate For the first function the argument gate_name is the name of the gate the message has to be sent through If this gate is a vector gate index determines though which particular output gate this has to be done otherwise the index argument is not needed The second function uses the gate number and because it does not have to search through the gate array it is faster than the first one Examples send new cMessage token out gate send new cMessage token vectorgate i int out_gate_id findGate out gate for i 0 i lt n itt send new cMessage packet out_gate_id wait in_time All message sending functions check that you actually own the message you are about to send If the message is with another module currently scheduled or in a queue etc youll get a runtime error The feature does not increase runtime overhead significantly because it uses the object ownership management it merely checks that the owner of the message is the module which wants to send it 6 4 2 Delayed sending It is often needed to model a delay processing time etc immediately followed by message sending In OMNeT 4 it is possible to implement it like this wait some_delay send msg outgate If the module needs to react to messages that arrive during the delay wait cannot be used and the timer mechanism described under Implementing timers would need to be employed
225. inter to a user supplied deallocation function a pointer to a user supplied duplication function and an item size If all three are O NULL no memory management is done that is the pointer is treated as a mere pointer This is the default behaviour If you supply only the item size and both function pointers are NULL cPar will use the delete operator to deallocate the memory area when the cPar object is destructed and it will use new char size followed by a memcpy to duplicate the memory area whenever the cPar object is duplicated If you need more sophisticated memory management you can supply your own deallocation and duplication functions All this is documented in detail in the class library reference An example for simple memory management double mem new double 15 cPar par par setPointerValue void mem par configPointer NULL NULL 15 sizeof double gt now if par goes out of scope it will delete the 15 double array The configPointer setting only affects what happens when the cPar is deleted duplicated or copied but does not apply to assigning new pointers That is if you assign a new void to the cPar you simply overwrite the pointer the block denoted by the old pointer is not deleted This fact can be used to extract some dynamically allocated block out of the cPar carrying on the previous example you would extract the array of 15 doubles from the cPar like this double mem2 double
226. ion With activity only The wait function s implementation contains a receive call which cannot be used in handleMessage The wait function suspends the execution of the module for a given amount of simulation time a delta regardless whether messages arrive at the module in the meanwhile or not wait delay_interval Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 104 In other simulation software wait is often called hold Internally the wait function is implemented by a scheduleAt followed by a receive The wait function is very convenient in modules that do not need to be prepared for arriving messages for example message generators An example for wait par interarrival time generate and send message The messages that arrived during the wait call will accumulate in the putaside queue The putaside queue can be examined directly an example was shown in the previous section and its contents is also retrieved by the receive or receiveOn functions 6 4 6 Self messages The module can send a message to itself using the scheduleAt function scheduleAt time msg scheduleAt accepts an absolute simulation time usually simTime something Messages sent via scheduleAt are called self messages and in OMNeT they are used to model events which occur within the module Self messages are del
227. ion s These are files with the ned suffix Simple modules These are C files with cc suffix Model files are usually placed in the projects modelname subdirectory of the main OMNeT directory The NED files are compiled into C using the NEDC compiler which is part of OMNeT The NEDC compiler source and executable is normally located in the nedc subdirectory of the main OMNeT directory The simulation system provides the following components that will be part of the simulation executable Simulation kernel with the simulation class library This is a library file with a or lib extension normally in the sim subdirectory of the main OMNeT directory It comes in two versions libsim_std a sim_std 1lib is the standard version and libsim_pvm a sim_pvm 1ib is the one to be used with parallel execution User interfaces These are also library files a or 1ib file normally in the envir directory and other directories The common part of all user interfaces is Libenvir a envir 1lib and the specific user interfaces are libcmdenv a cmdenv lib libtkenv a tkenv lib Simulation programs are built from the above components First the NED files are compiled into C source code using the NEDC compiler Then all C sources are compiled and linked with the simulation kernel and a user interface to form a simulation executable The following figure gives an overview of the process of building and running simulation pro
228. is provided for the following reason the cSimpleModule snapshot call outputs every object of the simulation into a text file which is excellent for debugging Unfortunately ordinary variables int char types etc do not appear in the snapshot file cWat ch helps this Use it like this int samples new cWatch samples samples Now the cWatch object will make the samples variable appear in the snapshot file The second line can be shortened using the WATCH macro WATCH samples Constructors cWatch cWatch amp vs cWatch const char name char amp c cWatch const char name int amp i cWatch const char name long amp 1 cWatch const char name double amp d Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 243 cWatch const char name char amp s cWatch const char name cObject amp 0 Initialize the shell to hold the given variable virtual const char className j virtual cObject dup All usual virtual functions redefined virtual void info char buf virtual void writeContents ostream amp os These functions are redefined to display the value of the variable Output is like this int samples 12 12U OxC virtual void printTo char buf Does actual work for info and writeContents 15 11 Output vector class cOutVector Header file coutvect h sim directory Th
229. it is at index g in the gate vector This function should not be directly called by the user Link parameters cLinkType link Returns the link type of the gate if it has one cPar delay cPar error cPar datarate Return pointers to the delay bit error rate and datarate parameters of the link Links are one directional these parameters are only stored at their starting side void setLink cLinkType 1 Sets the parameters of the link to those specified by the link type void setDelay cPar p void setError cPar p void setDataRate cPar p Set the parameters of the link Ownership of cPar objects are handled according to the ownership flag that is set by takeOwnership Topology bool isConnected Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 229 Returns t rue if the gate is connected cGate fromGate cGate toGate For a compound module gate it returns the previous and the next gate in the series of connections the route that contains this gate For simple module gates only one of the functions will return non NULL value void setFrom cGate g void setTo cGate g Redirect gates This function will rarely be needed unless maybe for dynamically created modules cGate sourceGate cGate destinationGate Return the ultimate source and destination of the series of connections the route that contains this
230. ivalent of OMNeT gates sendDelayed message gate delay hold delay wait delay Ses PEM iti receive msgtype msg msg receive delay or timeout after delay more complex receive constructs while msg receive if Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix C NED Language Grammar 206 14 APPENDIX C NED LANGUAGE GRAMMAR The NED language the network topology description language of OMNeT will be given using the extended BNF notation Space horizontal tab and new line characters counts as delimiters so one or more of them is required between two elements of the description which would otherwise be unseparable 7 two slashes may be used to write comments that last to the end of the line The language only distinguishes between lower and upper case letters in names but not in keywords In this description the xxx notation stands for one or more xxx s separated with spaces tabs or new line characters and xxx stands for one or more xxx s separated with a comma and optionally spaces tabs or new line characters For ease of reading in some cases we use textual definitions The networkdescription symbol is the sentence symbol of the grammar notation meaning a Oor 1 time a a a a 1 or more times a separated by commas a 1 or more times a separated by spaces alb aor b a the character a bold
231. ivered to the module in the same way as other messages via the usual receive calls or handleMessage the module may call the isSelfMessage member of any received message to determine if it is a self message Before self messages are delivered they can be cancelled removed from the FES This is particularly useful because self messages are often used to model timers cancelEvent msg The cancelEvent function takes a pointer to the message to be cancelled and also returns the same pointer After having it cancelled you may delete the message or reuse it in the next scheduleAt calls cancelEvent gives an error if the message is not in the FES The following example shows how to implement timers cMessage timeout_msg new cMessage scheduleAt simTime 10 0 timeout_msg Tlie cMessage msg receive if msg timeout_msg timeout expired else other message has arrived timer can be cancelled now delete cancelEvent timeout_msg You can determine if a message is currently in the FES by calling its isScheduled member if msg gt isScheduled delete cancelEvent msg else Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 105 An advanced version of the above code which also checks the put aside queue if msg gt isScheduled delete cancelEvent msg else if putAsideQueue contains msg
232. iversity of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 223 Removes the given message from the message queue The message needs to have been sent using the scheduleAt function This function can be used to cancel a timer implemented with scheduleAt Receiving new messages cMessage receiveNew simtime_t t MAXTIME Removes the next message from the event queue and returns a pointer to it If there is no message in the event queue the function waits with t timeout until a message will be available If the timeout expires and there is still no message in the queue the function returns NULL cMessage receiveNewOn const char s int sn 0 simtime_t t MAXTIME The same as receiveNew except that it returns the next message that arrives on the gate specified with its name and index All messages received meanwhile are inserted into the put aside queue If the timeout expires and there is still no such message in the queue the function returns NULL In order to process messages that may have been put in the put aside queue the user is expected to call receive or receiveOn or to examine the put aside queue directly sometime cMessage receiveNewOn int g simtime_t t MAXTIME Same as the previous function except that the gate must be specified with its index in the gate array Using this function instead the previous one may speed up the sim
233. jects can be reached through forEach on which many algorithms rely e g saveresults For cHead the dup and operator functions are NOT implemented dup would require that every object in the list be duplicated Since cHead is mostly an internal class and is NOT intended for use by the programmer as a container class the dup operation was considered unnecessary Construction destruction copying cHead const char name NULL cHead h amp locals bool init true Constructor Accepts the object name the head of the list it should join and a parameter that tells cHead if it should initialize the pointers to the list virtual cHead The destructor deletes all objects in the list that were created on the heap Redefined virtual functions virtual const char className j Returns the pointer to the class name string CcHead virtual void forEach ForeachFunc f Calls the function passed for each object in the list Member functions cObject find const char s Searches the list for an object with the given name and returns its pointer If no such object was found NULL is returned int count Returns the number of objects in the list Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 239 15 9 6 struct clterator To examine each element in the list the cIterator iterator class can be used Once a cIterator object is crea
234. l Tk dlls amp libs Instead of having the path hardwired the project files expect to find it in the TK_LIB_PATH environment variable so you have to set it before starting MSVC The names of the Tcl Tk libs is assumed to be tcl82 lib and tk82 lib if you use a different version you ll have to manually change the project files To build a sample simulation start MSVC and open the project dsp file The simulation should build without any project file adjustment To change from Tkenv to Cmdenv or vica versa choose BuildlSet active configuration from the menu and select one of Debug Tkenv Release Tkenv Debug Cmdenv Release Cmdenv then re link the executable If you have big models youll probably have to increase the stack size You ll find the setting under ProjectlSettings gt Link tab gt choose Output from combo gt Stack allocations Reserve Be aware that if you don t specify anything here MSVC defaults to 1MB way too small If you need to modify the names of the Tcl Tk libs because you installed a Tcl Tk version other than 8 2 see ProjectlSettings gt Link tab gt choose Input from combo gt Libraries The Tcl Tk install program normally sets the TCL_LIBRARY environment variable needed by Tcl applications However if you see the cant find a usable init tcl error message when you start a simulation program or Gned or Plove then that didn t happen and you ha
235. l dup the contained objects are duplicated too for the new container This is done so because contained objects are owned by the container ownership is defined as the right duty of deallocation However there is a fine grain ownership control mechanism built in which allows you to specify on per object basis whether you want objects to be owned by the container or not by calling the Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 95 takeOwnership member function with false you tell the container that you dont want it to become the owner of objects that will be inserted in the future You can explicitly select an owner for any object by calling its set Owner member function 6 2 Utilities Tracing The tracing feature will be used extensively in the code examples so it is shortly introduced here It will be covered in detail in a later section The ev object represents the user interface of the simulation You can send debugging output to ev with the C style output operators ev lt lt packet received sequence number is lt lt seq_num lt lt endl An alternative solution is ev printf ev printf packet received sequence number is d n seq_num Simulation time conversion There are utility functions which convert simulation time simtime_t to a printable string like 3s 130ms 230us and vica versa The simtimeToStr function converts a si
236. l gates are connected You can turn off this check by using the nocheck modifier This example generates a random subgraph of a full graph module Stochastic parameters gates submodules connections nocheck for i 0 9 do Sender outgate i gt Receiver i ingate if uniform 0 1 lt 0 3 endfor endmodule When using nocheck it is the simple modules responsibility not to send messages on gates that are not connected 4 6 Parameterized compound modules With the help of conditional parameter and gatesize blocks and conditional connections one can create complex topologies 4 6 1 Examples Example 1 Router The following example contains a router module with the number of ports taken as parameter The compound module is built using three module types Application RoutingModule DataLink We assume that their definition is in a separate NED file which we will import import modules module Router parameters rte_processing_delay rte_buffersize num_of_ports const gates in input_ports out output_ports Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 46 submodules local_user Application routing RoutingModule parameters processing_delay rte_processing_delay buffersi rte_buffersize gatesizes input num_of_ports l output num_of_ports 1 port_if DataLink num_of_ports parameters retry_count
237. lation Library 141 to_sink cGate gt sink in end detailed gate list deleted from here cHead token comp 0 mac local objects begin sendqueue length cOutVector single send queue cQueue n 11 end cOutVector token comp 0 mac local objects sendqueue length begin end cQueue token comp 0 mac local objects send queue begin 0 gt 1 cMessage Tarr 0 0158105774 15ms Src 4 Dest 3 0 gt 2 cMessage Tarr 0 0163553310 16ms Src 4 Dest 3 0 gt 1 cMessage Tarr 0 0205628236 20ms Src 4 Dest 3 0 gt 2 cMessage Tarr 0 0242203591 24ms Src 4 Dest 3 0 gt 2 cMessage Tarr 0 0300994268 30ms Src 4 Dest 3 0 gt 1 cMessage Tarr 0 0364005251 36ms Src 4 Dest 3 0 gt 1 cMessage Tarr 0 0370745702 37ms Src 4 Dest 3 0 gt 2 cMessage Tarr 0 0387984129 38ms Src 4 Dest 3 0 gt 1 cMessage Tarr 0 0457462493 45ms Src 4 Dest 3 0 gt 2 cMessage Tarr 0 0487308918 48ms Src 4 Dest 3 Q gt 2 cMessage Tarr 0 0514466766 51ms Src 4 Dest 3 end cMessage token comp 0 mac local objects send queue 0 gt 1 begin 4 gt 3 sent 0 0158105774 15ms arrived 0 0158105774 15ms length 33536 kind 0 priority error FALSE time stamp 0 0000000 0 00s parameter list dest cPar 1 L source cPar 0 L gentime cPar 0 0158106 D end
238. lation programmer has to code the algorithms often FSMs so that they can always properly respond to end of simulation events in whichever state they are This often makes program code unnecessarily complicated This fact is also evidenced in the design of the Parsec simulation language UCLA Its predecessor Maisie used end of simulation events but as documented in the Parsec manual this is led to awkward programming in many cases so for Parsec end of simulation events were dropped in favour of finish called finalize in Parsec 5 4 Finite State Machines in OMNeT Overview Finite State Machines FSMs can make life with handleMessage easier OMNeT provides a class and a set of macros to build FSMs OMNeT s FSMs work very much like OPNET s or SDL s The key points are Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 81 e There are two kinds of states transient and steady At each event that is at each call to handleMessage the FSM transitions out of the current steady state undergoes a series of state changes runs through a number of transient states and finally arrives at another steady state Thus between two events the system is always in one of the steady states Transient states are therefore not really a must they exist only to group actions to be taken during a transition in a convenient way e You can assign program code to entering a
239. lation run One might need to perform a large number of simulation runs where the model parameters are not known in advance This can be the case when one wants to optimize a system and parameter tuning cannot be used because 1 for each experiment he wants to start the model from a well defined initial state or 2 he wants to change the model topology from one simulation run to the other In this case the following solution be followed The network would consist of only one simple module that would organize the simulation runs by creating running and destroying the actual models with each experiment The simple module s code would look like this Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 91 SimulationManager activity determine parameters for the first run while true create the model a compound module with the current run parameters schedule wait some time while the model runs delete future events that belong to the model get Statistics out of the model destroy the model if simulation is done break calculate parameters for the next run write out results The solutions built into OMNeT flexible module topologies dynamic creation of compound modules etc strongly support this concept 5 8 5 Dynamic topology optimization Dynamic topology optimization is the generalization of the parameter tuning and multiple experiments within one simulation ru
240. lecommunications OMNeT Manual Simple Modules 71 another coroutine by a transferTo otherCoroutine call Then this coroutine is suspended and otherCoroutine will run Later when otherCoroutine does a_transferTo firstCoroutine call execution of the first coroutine will resume from the point of the transferTo otherCoroutine call The full state of the coroutine including local variables are preserved while the thread of execution is in another coroutines This implies that each coroutine must have an own processor stack and transferTo involves a switch from one processor stack to another Coroutines are at the heart of OMNeT 4 and the simulation programmer doesn ever need to call transferTo or other functions in the coroutine library nor does he need to care about the coroutine library implementation But it is important to understand how the event loop found in discrete event simulators works with coroutines When using coroutines the event loop looks like this simplified while FES not empty and simulation not yet complete retrieve first event from FES t timestamp of this event transferTo module containing the event That is when the module has an event the simulation kernel transfers the control to the module s coroutine It is expected that when the module decides it has finished the processing of the event it will transfer the control back to the simulation kernel by a transferTo main call Initially
241. line of code See the following example where Step 3 is omitted find descriptor object cModuleType moduleType findModuleType TCPConnectionHandler create possibly compound module and build its submodules if any cModule module moduleType gt create TCPconn this moduleType gt buildiInside module create activation message module gt scheduleStart simTime If you want to set up parameter values or gate vector sizes Step 3 the code goes between the create and buildiInside calls cModuleType moduleType findModuleType TCP conn handler cModule module moduleType gt create TCPconn this Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 112 set up parameters and gate sizes before we set up its submodules module gt par window size 4096 module gt setGateSize to apps 3 moduleType gt buildiInside module module gt scheduleStart simTime To delete a module dynamically module gt deleteModule If the module was a compound module this involves recursively destroying all its submodules A simple module can also delete itself in this case if the module was implemented using activity the deleteModule call does not return to the caller the reason is that deleting the module also deletes the CPU stack of the coroutine Currently you
242. llection is in the first phase transforms the values into histogram If already in the second phase the function does nothing virtual unsigned cells Returns the number of cells virtual double basepoint unsigned k Returns the kth basepoint virtual unsigned cell unsigned k Returns the number of samples that fell into the kth subinterval virtual double cellPDF unsigned k Returns the calculated PDF in the kth subinterval virtual double random Returns a random number based on the distribution collected If no values have been collected it returns 0 when in initial collection phase it returns one of the stored observations after the histogram has been set up a random integer is returned virtual void clearResult Clears all results collected so far 15 12 9 class cDoubleHistogram cDoubleHistogram is derived from cEqdHistogramBase see that for more information Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 252 Construction destruction copying cDoubleHistogram cDoubleHistogram amp r Copy constructor cDoubleHistogram const char s NULL unsigned int num_cells 10 Constructor that takes the object name the number of subintervals and the range factor as argument virtual cDoubleHistogram Destructor virtual cObject dup cDoubleHistogram amp operator cDoubleHis
243. llection object cDoubleHistogram hist the distribution cPar rnd2 rnd2 rnd2 setDoubleValue hist 6 12 3 Storing object and non object pointers in cPar cPar can store pointers to OMNeT objects You can use both assignment and the setObjectValue member function cQueue queue new cQueue queue just an example cPar parl par2 parl cObject queue par2 setObjectValue queue To get the store pointer back you can use typecast or the ob ject Value member function Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 123 cQueue ql cQueue cObject parl1 cQueue q2 cQueue par2 objectValue Whether the cPar object will own the other object or not is controlled by the takeOwnership member function just as with container classes This is documented in detail in the class library reference By default cPar will own the object cPar can be used to store non object pointers for example C structs or non OMNeT object types in the parameter object It works very similarly to the above mechanism An example double mem new double 15 cPar parl par2 parl void mem par2 setPointerValue void mem double m1 double void parl double m2 double par2 pointerValue Memory management can be specified by cPar s configPointer member function It takes three arguments a po
244. lls in a single printf statement ev printf objectl is s object2 is s n object1 gt fullPath object 2 gt fullPath yee WRONG Same string is printed twice Copying and duplicating objects The dup member function creates an exact copy of the object duplicating contained objects also if necessary This is especially useful in the case of message objects dup returns a pointer of type cObject so it needs to be cast to the proper type cMessage copyMsg cMessage msg gt dup dup works through calling the copy constructor which in turn relies on the assignment operator between objects operator can be used to copy contents of an object into another object of the same type The copying is done properly object contained in the object will also be duplicated if necessary For various reasons operator does not copy the name string the copy constructor does it Iterators There are several container classes in the library cQueue cArray etc For many of them there is a corresponding iterator class that you can use to loop through the objects stored in the container For example cQueue queue Ll kee for cQueuelterator queuelter queue queuelIter end queuelter t cObject containedObject queuelter Ownership control By default if a container object is destroyed it destroys the contained objects too If you cal
245. lso supported OMNeT provides several random number generators and also a tool for selecting good seed values Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix A OPNET and OMNeT 190 Process description method Method finite state machine both process style graphical spec only coroutine based and finite state machine textual spec only Direct non scheduled process interaction E Ce OMNeT o O eree ae interrupt Daa function call of other module Dynamic module creation What can be created only processes within an simple modules existing module connections compound modules with arbitrarily complex parameterized topologies Object oriented concepts Objects C API functions operating full flexibility of C on object like data inheritance polymorphism structures etc no support for inheritance built in object oriented polymorphism or the like mechanisms The graphical user interface of OPNET from version 3 0 contains an inheritance mechanism for models This is no real inheritance in the object oriented sense because it just means that parameter values can be changed or fixed down parameters renamed merged etc There is no mention about changing the behaviour of a module that is anything like C s virtual functions Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix A OPNET and OMNeT 19
246. lues to nodetype such as WormholeRoutingNode DeflectionRoutingNode etc Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 51 4 7 Network definition A network definition or system module definition specifies the system module In its syntax it is very similar to a submodule declaration The system definition starts with keyword network and ends with endnetwork An example network modelledNetwork SomeModule parameters parl 10 par2 normal 100 20 endnetwork Here SomeModul1e is the name of a compound or a simple module type There can be several system definitions in a network description each one defines a different network The simulation program built with such a network description is able to run any of them the desired one can be specified in the config file see later 4 8 Support for parallel execution OMNeT simulations can be executed in parallel This means that different parts of the model execute on different hosts or processors We ll use the term host or machine in this sense The unit of granularity is the simple module one simple module always executes on a single processor Parallel execution is also supported by NED the language provides an elegant way of specifying execution hosts for different modules We ll discuss this feature in the following sections 4 8 1 Extensions to the compound module and system definitions To support th
247. make clean make 7 3 Using Win32 with Borland C 5 0 7 3 1 Compiling the libraries and nedc IDE files are provided for all OMNeT libraries Sim Envir Cmdenv Tkenv You should be able to build the lib files from the Borland C IDE without problem You may need to adjust the include and lib paths in Options Project gt Directories Tkenv needs special treatment You need to get the Tcl and Tk sources I tried with 8 0 patchlevel 2 tcl80p2 zip and tk80p2 zip build the libraries and install them properly Tkenv includes the Tk header files so before compiling Tkenv you may need to adjust the paths in Options Project gt Directories The default location of Tkenv s TCL script parts and the bitmaps directory is passed to the Tkenv as a compile time external define At run time they can be overridden be the OMNETPP_TKENV_DIR and OMNETPP_ BITMAP PATH environment variables You ll probably need to adjust those paths too they are in Options Project gt Compiler gt Defines To compile nedc you ll need to get a bison flex pair that works in the Win95 NT command box Alternatively you can compile it in DOS with Borland C 3 1 and with bison flex from the DJGPP package Technical University of Budapest Dept of Telecommunications OMNeT Manual Building Simulation Programs 151 7 3 2 Setting up a project file It is best starting by copying an example simulation s project file to a different name and modify th
248. manuals Word6 HTML readmes license etc system description compiler simulation kernel files for non distributed execution files for distributed execution over PVM common code for user interfaces command line user interfac Tcl Tk based user interfac graphical NED editor output vector analyzer and plotting tool makefile autocreator etc are included The cont rib directory contains material from the OMNeT community omnetpp samples nim hcube token fddi hist dyna pvmex fifol ifo2 demo contrib octave emacs directories for sample simulations a simple two player game hypercube network with deflection routing Token Ring network an accurate FDDI MAC simulation demo of the histogram classes dynamic module creation client server network demonstrates distributed execution sing server queu another implementation of a single server queu several sim models in a single executable directory for contributed material Octave scripts for result processing NED syntax highlight for Emacs Technical University of Budapest Dept of Telecommunications OMNeT Manual An Example The NIM Game 27 3 AN EXAMPLE THE NIM GAME This chapter contains a full example program that can give you some basic idea of using the simulator An enhanced version of the NIM example can be found among the sample programs Nim i
249. me gt Specify the name of the configuration file The default is omnetpp ini Multiple f switches can be given this allows you to partition your configuration file For example one file can contain your general settings another one most of the module parameters another one the module parameters you change often l lt fileName gt Load a shared object so file on Unix Multiple 1 switches are accepted Your So files may contain module code etc By dynamically loading all simple module code and compiled network description _n o files on Unix you can even eliminate the need to re link the simulation program after each change in a source file Shared objects can be created with gcc shared r lt runs gt Only recognized by simulations linked with Cmdenv It specifies which runs should be executed e g r2 4 6 8 This option overrides the runs to execute option in the Cmdenv section of the ini file see later All other options are read from the configuration file An example of running an OMNeT executable with the h flag C OMNETPP PROJECTS FDDI gt fddi exe h OMNeT Discrete Simulation TUB Dept of Telecommunications 1990 97 Networks in this program 1 NRing 2 FDDI1 End run of OMNeT 8 2 The configuration file omnetpp ini 8 2 1 Sections and entries The configuration file also called ini file because it has an ini extension contains options that control how the s
250. mt ime_t passed in the first arg to textual form The result is placed into the buffer pointed to by the second arg If the second arg is omitted or it is NULL simtimeToStr will place the result into a static buffer which is overwritten with each call char buf 32 ev printf t1l s t2 s n simtimeToStr t1 simTimeToStr t2 buf The strToSimtime function parses a time specification passed in a string and returns a simtime_t If the string cannot be entirely interpreted 1 is returned simtime_t t strToSimtime 30s 152ms Another variant st rToSimtime0 can be used if the time string is a substring in a larger string Instead of taking a char it takes a reference to char char amp as the first argument The function sets the pointer to the first character that could not be interpreted as part of the time string and returns the value It never returns 1 if nothing at the beginning of the string looked like simulation time it returns 0 const char s 30s 152ms and some rubbish simtime_t t strToSimtime0 s now s points to and some rubbish Utility lt string h gt functions The opp_strdup opp_strcpy opp_strcemp functions are the same as their lt string h gt equivalents except that they treat NULL and the empty string as identical and opp_strdup uses operator new instead of malloc Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simula
251. n Parameters taken by reference can be used as a second means of module communication because during simulation execution if a module changes the value of a parameter taken be reference the Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 42 changed value propagates to other modules ref parameters can also be used to implement shared memory see in Chapter 5 Submodule gate sizes The sizes of gate vectors are defined with the gatesizes keyword Gate vector sizes can be given as constants parameters or expressions An example simple SimpleType gates in inputs out outputs endsimple module SomeCompound parameters num const submodules Submod1l SimpleType gatesizes inputs 10 outputs num Pf ies endmodule Conditional parameter and gatesize sections Multiple parameters and gatesizes sections can exist in a submodule definition and each of them can be tagged with conditions For example module Serial parameters count const submodules node Node count parameters position middle parameters if index 0 position beginning parameters if index count 1 position end gatesizes in 2 out 2 gatesizes if index 0 index count l in l in 1 connections eee endmodule If the conditions are not disjunct and a parameter value or a gate size is defined twice the last definition will take effect overwriting the
252. n new sXElem 3 expression 0 amp par mean expression 1l amp par stddev expression 2 normal pointer to the normal double double func cPar expr expr expr setDoubleValue expression 3 For more information see the reference and the code NEDC generates for parameter expressions 6 12 5 Using redirection A cPar object can be set to stand for a value actually stored in another cPar object This is called indirect or redirected value When using redirection every operation on the value i e reading or changing it will be actually done to the other cPar object Operations on A asking the value changing the value cPar A redirected to B Y eos cPar B actually done on B ordinary value Fig 5 3 cPar redirection Redirection is how module parameters taken by reference are implemented The redirection does not include name strings That is if you say A gt setName newname in the above example A s name will be changed as the name member is not redirected This is natural if you consider parameters taken by reference a parameter should can have different name than the value it refers to You create a redirection with the setRedirection function cPar bb new cPar bb background value Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 125 bb 10L cPar a a we ll redirect this object a setRedirection bb
253. n Then the embedding program will take the place of Cmdenv and Tkenv the only exception is textual module output it is sent directly to ENVIR ev lt lt hello ev print world Technical University of Budapest Dept of Telecommunications OMNeT Manual The Design of OMNeT 181 If ENVIR does not fit your needs for example you want the model parameters to come from a database not from omnetpp ini then you have replace it Your ENVIR replacement the embedding program practically must implement the cEnvir member functions from envir cenvir h but you have full control over the simulation Normally code that sets up a network or builds the internals of a compound module comes from compiled NED source You may not like the restriction that your simulation program can only simulate networks whose setup code is linked in No problem your program can contain pieces of code like what is generated by nedc and then it can build any network whose components primarily the simple modules are linked in It is even possible to write an integrated environment where you can put together a network using a graphical editor and right after that you can run it without intervening NED compilation and linkage 11 3 The simulation kernel The source code for the simulation kernel of OMNeT and the library classes reside in the sim directory Almost all objects are derived from cOb ject which provides a common interface for them 11 3
254. n concepts If one wants to simulate large systems it is possible that one part of the model needs its topology to be optimized optimal number of servers optimal interconnection etc while other parts of the model have reached their steady state and should not be bothered This can be achieved by modifying the previous scheme Parts of the model that do not need topology optimization can be created once and left running for the whole duration of the simulation other parts are examined and their structure is modified from time to time Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 92 6 THE SIMULATION LIBRARY OMNeT has a rich C class library which you can use when implementing simple modules A quick overview of the areas supported by the simulation library e sending and receiving messages scheduling and cancelling events terminating the module or the simulation member functions of cSimpleModule e events messages network packets the cMessage and cPacket classes e random number generation normal exponential etc e access to module gates and parameters member functions of cModule base class for cSimpleModule cPar and cGate classes e accessing other modules in the network member functions of cModule and cGate e storing data in containers cArray cQueue cBag and cLinkedList classes e discovering network topology and support routing cTopology class e re
255. n Library 108 Link parameters The following member functions return the link attributes cLinkType link outgate gt link cPar d outgate gt delay cPar e outgate gt error cPar r outgate gt datarate Transmission state The isBusy member function returns whether the gate is currently transmitting and if so the transmissionFinishes member function returns when it will finish transmitting Connectivity TBD figure The isConnected member function returns whether the gate is connected If the gate is an output gate the gate to which it is connected is obtained by the toGate member function For input gates the function is fromGate cGate gate gate somegate if gate gt isConnected cGate othergate gate gt type 0 gate gt toGate gate gt fromGate ev lt lt gate is connected to lt lt othergate gt fullPath lt lt endl else ev lt lt gate not connected lt lt endl An alternative to isConnected is to check the return value of toGate or fromGate The following code is fully equivalent to the one above cGate gate gate somegate cGate othergate gate gt type 0 gate gt toGate gate gt fromGate if othergate ev lt lt gate is connected to lt lt othergate gt fullPath lt lt endl else ev lt lt gate not connected lt lt endl To find out to which simple modul
256. n as intuniform 1 2 truncnormal is the normal distribution truncated to nonnegative values its implementation generates a number with normal distribution and if the result is negative it keeps generating other numbers until the outcome is nonnegative The counterparts of the above functions using generator k double genk_uniform double k double lower_limit double upper_limit double genk_intuniform double k double lower_limit double upper_limit double genk_exponential double k double mean double genk_normal double k double mean double deviation double genk_truncnormal double k double mean double deviation Note that they take the number of the generator as a double it is so because these functions are designed so that they can be used with the cPar class and in NED files You will find more information about this in the section describing cPar Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 118 If the above distributions do not suffice you can write your own functions If you register your functions with the Register_Function macro you can use them in NED files and ini files too You can find the implementation of many distributions in the class library of GNU C 6 9 3 Random numbers from histograms You can also specify your distribution as a histogram The cLongHistogram cDoubleHistogram cVarHistogram cKSplit or cPSquare
257. n it with a simple and fast user interface that supports batch execution Component libraries Module types can be stored in files separately from the place of their actual usage This means that the user can group existing module types and create component libraries Universal standalone simulation programs A simulation executable can store several independent models that use the same set of simple modules The user can specify in the configuration file which model he she wants to run This allows one to build one large executable that contains several simulation models and distribute it as a standalone simulation tool The flexibility of the topology description language also supports this approach Technical University of Budapest Dept of Telecommunications OMNeT Manual Overview 2 3 2 What is what in the directories To help you navigate among files in the OMNeT distribution here s a list what you can find in the different directories The omnet pp directory contains the following subdirectories The simulation system itself omnetpp bin lib doc src nedc sim std pvm envir cmdenv tkenv gned plove utils Sample simulations are within the samples directory Each of the sample directories contain a network description ned file and corresponding simple module code h cc files Makefiles OMNeT root directory soft links to executables in src soft links to sim library files in src
258. n simple modules with the randseed function randseed 10 set seed to 10 long seed randseed current seed value Zero is not allowed as a seed The intrand and randseed functions use generator 0 They have another variant which uses a specified generator long rnd genk_intrand 6 like intrand using generator 6 genk_randseed k 167 set seed of generator k to 167 The intrand n and dblrand functions are based on intrand int dice 1 intrand 6 result of intrand 6 is in the range 0 5 it is calculated as intrand 6 double prob dblrand in the range 0 0 1 0 calculated as intrand double INTRAND_MAX They also have their counterparts that use generator k int dice 1 genk_intrand k 6 uses generator k double prob genk_dblrand k uu 6 9 2 Random numbers from distributions The following functions are based on dblrand and return random variables of different distributions doub uniform double lower_limit double upper_limit Le double intuniform double lower_limit double upper_limit double exponential double mean le le doub normal double mean double deviation double truncnormal double mean double deviation They are the same functions that can be used in NED files intuniform generates integers including both the lower and upper limit so for example the outcome of tossing a coin could be writte
259. n the evaluation stack The cPar which evaluates this expression will copy the cPar for itself void operator char _op The argument can be add subtract multiply divide Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 234 modulo power of equal not equal greater greater or equal less less or equal inline if C language s cond a b Effect during evaluation of the expression two items or three with 2 are popped out of the stack the given operator is applied to them and the result is pushed back on the stack void operator void operator thFuncNoArg f thFunclArg f void operator thFunc2Args f void operator thFunc3Args f The argument can be a pointer to a function that takes 0 1 2 or 3 double arguments and returns a double e g sqrt Effect during evaluation of the expression the given number of doubles are popped from the stack the given function is called with them as arguments and the return value is pushed back on the stack See also the cFunctionType class and the Define _Function macro Ma Ma Ma Ma The OMNeT functions generating random variables of different distributions can also be used in sXElem expressions 15 9 Object container classes cArray cCQueue cHead Header files carray h cqueue h sim directory There are four container classes cArray and cQueue cHead
260. nd leaving a state known as entry exit code Staying in the same state is handled as leaving and re entering the state e Entry code should not modify the state this is verified by OMNeT State changes transitions must be put into the exit code OMNeT s FSMs can be nested This means that any state or rather its entry or exit code may contain a further full fledged FSM_Switch see below This allows you to introduce sub states and thereby bring some structure into the state space if it would become too large The FSM API FSM state is stored in an object of type cF SM The possible states are defined by an enum the enum is also a place to tell which state is transient and which is steady In the following example SLEEP and ACTIVE are steady states and SEND is transient the numbers in parens must be unique within the state type and they are used for constructing the numeric IDs for the states enum INIT 0 SLEEP FSM_Steady 1 ACTIVE FSM _Steady 2 SEND FSM_Transient l1 The actual FSM is embedded in a switch like statement FSM_Switch where you have cases for entering and leaving each state FSM_Switch fsm case FSM _Exit statel beh break case FSM_Enter statel Tp aia break case FSM Exit state2 Id iu break case FSM_Enter state2 LA wars break LY ex H State transitions are done via calls to FSM_Goto whic
261. nd should be redefined in descendants Getting the current simulation time simtime_t simTime Returns the current simulation time that is the arrival time of the last message returned by a receiveNew call Sending messages int send cMessage msg int g Sends a message through the gate given with its ID int send cMessage msg const char s int sn 0 Sends a message through the gate given with its name and index if multiple gate int send cMessage msg cGate outputgate Sends a message through the gate given with its pointer int sendDelayed cMessage msg double delay int g Sends a message through the gate given with its index as if it was sent delay seconds later int sendDelayed cMessage msg double delay const char s int sn 0 Sends a message through the gate given with its name and index if multiple gate as if it was sent delay seconds later int sendDelayed cMessage msg double delay cGate outputgate Sends a message through the gate given with its pointer as if it was sent delay seconds later int sendDirect cMessage msg double delay cGate inputgate TBD also to include other sendDirect funcs int scheduleAt simtime_t t cMessage amp msg Inserts the given message into the Future Event Set and schedules it to be returned at time t This function can be used to implement timers cMessage cancelEvent cMessage msg Technical Un
262. ne FSM_DEBUG include lt omnetpp h gt class BurstyGenerator public cSimpleModule public Module_Class_Members BurstyGenerator cSimpleModule 0 parameters double sleepTimeMean double burstTimeMean double sendIATime cPar msgLength FSM and its states cFSM fsm enum INIT 0 SLEEP FSM_Steady 1 ACTIVE FSM_Steady 2 SEND FSM_Transient 1 F variables used int i cMessage startStopBurst cMessage sendMessage Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules the virtual functions virtual void initialize virtual void handleMessag Define Module void BurstyGenerator initiali BurstyGenerator fsm setName fsm cMessag e sleepTimeMean par sleep_time_mean burstTimeMean par burst_time_mean sendIATime par send_ia_time 83 msg msgLength amp par msg_length i 0 WATCH i always put watches in initialize startStopBurst new cMessage StartStopBurst sendMessage new cMessage sendMessage scheduleAt 0 0 startStopBurst void BurstyGenerator handleMessage cMessage msg FSM_Switch fsm case FSM_Exit INIT transition to SLEEP state FSM_Goto fsm SLEEP break
263. ned char amp c inline cEnvir amp operator gt gt cEnvir amp ev short i inline cEnvir amp operator gt gt cEnvir amp ev int amp i inline cEnvir amp operator gt gt cEnvir amp ev long amp 1 inline cEnvir amp operator gt gt cEnvir amp ev unsigned short amp i inline cEnvir operator gt gt cEnvir amp ev unsigned int i inline cEnvir amp operator gt gt cEnvir amp ev unsigned long amp 1 inline cEnvir amp operator gt gt cEnvir amp ev float f inline cEnvir amp operator gt gt cEnvir amp ev double amp d inline cEnvir amp operator gt gt cEnvir amp ev long double amp d Input operators for all fundamental data types But beware each gt gt operator reads a whole line 15 18 Classes not described in this reference cSimulation A class that holds the module tree of the network being simulated It has only one instance simulation cModuleType Class for simple module types cLinkType Class for channel types Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 256 cFunctionType Class for function types Used by the code generated by NEDC cNetworkType Class for pre compiled networks cClassRegister Class for registering a class Technical University of Budapest Dept of Telecommunications OMNeT Manual References 257 16 REFERENCES Simulation related JAINO1
264. nerated and additional insight can be obtained by applying mathematical operators to the collected data OPNET provides an advanced animation capability for visualising simulation events Both automatic and user customised animations can be displayed interactively during or after a simulation Animations can depict messages flowing between objects control flow in a process paths of mobile nodes and dynamic values such as queue size or resource status OPNET provides open system features including interfaces to standard languages the ability to take advantage of third party libraries an application program interface access to databases and data files such as those generated by network analysers and PostScript and TIFF export for desktop publishing OPNET users are guided by a comprehensive documentation set and are backed by outstanding technical support OPNET is very well designed and built commercial simulation software The author of OMNeT has worked for the Hungarian distributor of OPNET for over three years and he has gained significant experience with the software He has taken part in several computer network simulation projects for major Hungarian companies and also delivered OPNET training He has also written simulation models for a VSAT system in OPNET Following is a comparison of the features that concern general purpose computer systems simulation and are not specific to computer network simulation and that are present b
265. ng and its getter counterpart is named something i e the get verb found in Java and some other libraries is omitted for brevity For example the length attribute of the cMes sage class can be set and read like this msg gt setLength 1024 length msg gt length l4 className For each class the className member function returns the class name as a string const char classname msg gt className returns cMessage Name attribute An object can be assigned a name a character string The name string is the first argument to the constructor of every class and it defaults to NULL no name string If you supply a name string the object will make its own copy st rdup As an example you can create a message object like this cMessage mymsg new cMessage mymsg You can also set the name after the object has been created mymsg gt setName mymsg You can get a pointer to the internally stored copy of the name string like this const char name mymsg gt name gt returns ptr to internal copy fF of mymsg For convenience and efficiency reasons the empty string and NULL are treated as equivalent by library objects is stored as NULL so that it does not consume heap but it is returned as so that it is easier to print out etc Thus if you create a message object with either NULL or as name it will be stored as NULL and name will return a poin
266. ng module as an ellipse with the given height and width Defaults width height are chosen automatically Specifies options for the rectangle or oval Any valid Tk color specification is accepted English color names or reb rrggbb format where r g b are hex digits Defaults fillcolor 8080ff a lightblue outlinecolor black borderwidth 2 Meaning Drawing mode Specifies the exact placement of the connection arrow The arguments can be abbreviated as a e w n S m manual srcpx srcpy destpx destpy o color width Examples m a o blue 3 The manual mode takes four parameters that explicitly specify anchoring of the ends of the arrow srcpx srcpy destpx destpy Each value is a percentage of the width height of the source destination module s enclosing rectangle with the upper left corner being the origin Thus m m 50 50 50 50 would connect the centers of the two module rectangles Specifies the appearance of the arrow Any valid Tk color specification is accepted English color names or rgb rrggbb specification where r g b are hex digits Defaults color black width 2 4 10 GNED Graphical NED Editor The GNED editor allows you to design compound modules graphically GNED works with NED files it doesn t use any nasty internal file format You can load any of your existing NED files edit the compound modules in it graphically and then save the file back The rest of the
267. ng simulation results 6 14 1 Output vectors cOutVector Objects of type cOutVector are responsible for writing time series data referred to as output vectors to a file The record member is used to output a value or a value pair with a timestamp It can be used like this cOutVector resp_v response time while double response_time If ere resp_v record response_time Tf sex All cOut Vector objects write to the same common file The file is textual each record call generates a line in the file The output file can be processed using Plove but otherwise its simple format allows it to be easily processed with sed awk grep and the like and it can be imported by spreadsheet programs The file format is described later in this manual in the section about simulation execution You can disable the output vector or specify a simulation time interval for recording either from the ini file or directly from program code cOutVector v v simtime_t t v enable v disable v setStartTime t v setStopTime t 100 0 Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 135 If the output vector object is disabled or the simulation time is outside the specified interval record doesnt write anything to the output file However if you have a Tkenv inspector window open for the output vector object the values will be displayed
268. nications OMNeT Manual The Simulation Library 145 The above simple module code models the message generation part of a computer on a LAN The module sends out messages packets to different stations in every 1 5 seconds or so The messages have three parameters the source address the destination address and a sequence number The source address is the same in each packet and there are only three possible destination stations The sequence number is different in each packet To avoid the overhead caused by having to create source and destination address objects for each message the module creates these objects only once they will be shared among all messages Separate sequence number objects are created for each message Let us see what happens to the sequence number object when it is inserted into the message The message object by default takes the ownership of the object Ownership means the responsibility of destruction that is when the message is destroyed the parameter object will be destroyed as well This is exactly what we need most of the time But if we just added the shared source and destination address objects to a message then we would have problems when the message is destroyed Somehow it must be told to the message object to leave our shared parameters alone and not to become their owner This is exactly what the packet gt parList takeOwnership false line does it sets a flag that tells the message t
269. no underscores sendUnnumberedFrame On parameter names you may use C style window_size or C 4 style windowSize naming whichever you prefer Make the functions virtual Maybe someone who reuses your code will need a different behavior than what you thought of Use inheritance if you re writing a very complex simple module create a basic simple module class and build upon it deriving new module classes This will make your code more readable and easier to manage reuse Unfortunately inheritance is not supported in NED so you actually have to make distinct NED descriptions for each simple module class Even if you have an abstract classes prepare a NED description for it it is useful as a reference to others who might derive a different simple module class from your abstract class Inheritance in NED is planned in later releases of OMNeT Avoid global variables and what s the same static class members They are not reset to their initial value zero when you run the simulation stop it and rebuild the network This can cause several problems when you use Cmdenv to execute several runs one after another or in Tkenv when you rebuild the network from the menu Query the values of parameters into state variables gt class members of the same name at the top of the activity function If you know the value of a parameter is a random value like uniform 0 10 or it can change during simulation then to avoid having to look it up b
270. ns OMNeT Manual Appendix D Class Library Reference 217 different processors using the statistical synchronization method Currently this is implemented over PVM Parallel Virtual Machine version 3 virtual int netPack const char type NULL virtual int netUnpack These functions are expected to be redefined in all derived objects In OMNeT s PVM interface they call pvm_pkint ovm_upkint etc functions Functions supporting snapshots virtual void writeTo ostream amp os This function is called internally by cSimpleModule snapshot It writes out info about the object into the stream Relies on writeContents writeTo does not need to be redefined virtual void writeContents ostream amp os This function is called by internally writeTo It is expected to write textual information about the object and other objects it contains to the stream The default version cObject writeContents uses forEach to call info for contained objects Redefine as needed Other functions static int cmpbyname cObject one cObject other This function compares to objects by name It can be used in a priority queue class cQueue as a sorting criterion The forEach mechanism virtual void forEach ForeachFunc f Makes sense with container objects derived from cOb ject Calls the f function recursively for each object contained in this object The forEach mechanism The forEach mechanism implem
271. nsistent object oriented design One can freely use OOP concepts inheritance polymorphism etc to extend the functionality of the simulator Elements of the simulation messages modules queues etc are represented as objects These classes are part of the simulation class library e modules gates connections etc e parameters e messages e container classes e g queue array e data collection classes e statistic and distribution estimation classes histograms P algorithm for calculating quantiles etc e transient detection and result accuracy detection classes The objects are designed so that they can efficiently work together creating a powerful framework for simulation programming 2 2 1 Creating simple modules Each simple module type is implemented with a C class Simple module classes are derived from a simple module base class by redefining the virtual function that contains the algorithm The user can add other member functions to the class to split up a complex algorithm he can also add data members to the class It is also possible to derive new simple module classes from existing ones For example if one wants to experiment with retransmission timeout schemes in a transport protocol he can implement the protocol in one class create a virtual function for the retransmission algorithm and then derive a family of classes that implement concrete schemes This concept is further supported by the fact that in the
272. nstructor cStatistic const char s NULL Constructor creates an object with the given name virtual cStatistic The destructor does nothing cStatistic amp operator cStatistic amp res The assignment operator is present since descendants may refer to it Redefined virtual functions virtual const char className Returns a pointer to the class name string cStatistic Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 247 New non virtual member functions void setGenK int gen_nr Sets the index of the random number generator to use when the object has to generate a random number based on the statistics stored void addTransientDetection cTransientDetection object void addAccuracyDetection cAccuracyDetection object Assigns transient and accuracy detection objects to the statistic object cTransientDetection transientDetectionObject cAccuracyDetection accuracyDetectionObject Returns the assigned transient and accuracy detection objects New virtual member functions These pure virtual functions provide a common interface to all cStat istic classes void collect double val Collect one value virtual long samples j virtual double min virtual double max virtual double mean virtual double variance virtual double stddev virtual double sum virtual double sqrSum
273. nt and result accuracy detection sssessesssessseeesssetsseesseessersseressseesseessesseessees 253 15 14 11 class cTr ansientDetectoin k ce e e E E E aaeine 253 15 14 2 class cTDExpandingWindows sseseeeeeesseseeseessesresseesresressersresreesreseesrreseesese 253 15 14 3class CACCUrACY Dele COM kei a san tet acid a a 253 PS TAA class cADByStddev inaen a e Pah eae was E E E E 253 15 15 Finite State Machine CPS M ysis scscsiiccvasacaisa dca aari niia apai ei iieis 253 15 16 Routine s pport c Topilo yasina e TE contents 253 15 17 The user interface cEnvir and EV p issnscsstsi itsosue s gesoet 253 15 17 Velass CENY crie ane R a E EE 254 15 18 Classes not described in this reference seessesssesssesessseessresserssersseresssesssessersseesseee 255 Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 16 REFERENCES csccielccdendseuciccuciceseaeceicdedeckuslowducsesiiusisduiseasGeiveuwiae auedeawatassbibeatosmeatesawale Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 14 1 INTRODUCTION 1 1 What is OMNeT OMNeT is an object oriented modular discrete event simulator The name itself stands for Objective Modular Network Testbed in C OMNeT has its distant roots in OMNeT a simulator written in Object Pascal by dr Gy rgy Pongor The simulator can be used for modeling e communication protocols e computer networks and tra
274. nt at a data structure kept by the module which can carry enough context information about the event 6 3 5 The cPacket class The cPacket class is derived from cMessage It is intended as a base for all messages that model packets or frames in a telecommunications network cPacket adds two new data members to cMessage protocol and PDU type packet frame event type Both are short integers and are handled by the following member functions short protocol short pdu setProtocol short p setPdu short p Acceptable message kind values are e MK_PACKET e MK_INFO The cPacket constructor sets the message kind to MK_PACKET Both MK_PACKET and MK_INFO are defined as negative integers Remember negative message kind values are reserved for the simulation library The protocol and PDU fields would ideally take a value from the protocol h header in the simulation library The contents of protocol h is currently experimental comments and contributions are welcome TDB examples for protocol and pdu values 6 3 6 Subclassing cMessage and cPacket TBD include an example Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 101 6 4 Sending and receiving messages 6 4 1 Sending messages Once the message has been created it can be sent through an output gate using one of these functions send cMessage msg const char gate_name int index send cMessage msg
275. nt stage of a simulation or for presentation and educational purposes since it allows one to get a detailed picture of the state of simulation at any point of execution and to follow what happens inside the network The most important feaures are e message flow animation e graphical display of statistics histograms etc and output vectors during simulation execution e separate window for each module s text output e scheduled messages can be watched in a window as simulation progresses e event by event normal and fast execution e labeled breakpoints e inspector windows to examine and alter objects and variables in the model e simulation can be restarted e snapshots detailed report about the model objects variables etc Tkenv makes it possible to view simulation results output vectors etc during execution Results can be displayed as histograms and time series diagrams This can speed up the process of verifying the correct operation of the simulation program and provides a good environment for experimenting with the model during execution When used together with gdb or xxgdb Tkenv can speed up debugging a lot Portability Tkenv is built with Tcl Tk Tkenv should work on all platforms that Tcl Tk has been ported to Unix X Win32 Macintosh You can get more information about Tcl Tk in the Web pages listed in the Reference Simulation running modes in Tkenv Tkenv has the following modes for running the simulation e
276. o the user allow him her to start stop simulation execution and to intervene by changing variables objects inside the model This is very important in the development debugging setPhase of the simulation project User interfaces also facilitate demonstration of how a model works Since it was written in C the simulator is basically portable it should run on most platforms with a C compiler OMNeT s advanced user interfaces support X window DOS and are portable to Win3 1 Win95 WinNT OMNeT has been extended to execute the simulation in parallel Any kind of synchronization mechanism can be used One suitable synchronization mechanism is the statistical synchronization for which OMNeT provides explicit support OMNeT Home Page on the Web http www hit bme hu phd vargaa omnetpp htm Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 15 1 2 Where is OMNeT in the world of simulation tools There are numerous network simulation tools on the market today both commercial and non commercial In this section I will try to give an overview by picking some of the most important or most representative ones in both categories and comparing them to OMNeT Parsec SMURPH NS Ptolemy NetSim C SIM CLASS as non commercial and OPNET COMNET III as commercial tools The OMNeT Home Page contains a list of Web sites with collections of references to network simulation tools where the reader
277. o be more precise to its internal parameter list object not to take the ownership of objects that will be inserted from then on It does not affect objects already inserted As a result all messages will just hold pointers to the shared cPar objects and never do any harm to them The above example shows that with CPU intensive simulations you can save a lot of computation time and memory just by using the ownership mechanism already present in OMNeT 6 19 Building large networks There are situations when using NED files to describe network topology is inconvenient for example because the topology information comes from an external source e g it is exported from a network management program In such case you have two possibilities to avoid writing NED files by hand 1 generating NED files from data files 2 building the network from C code The two solutions have different advantages and disadvantages The first is more useful in the model development phase while the second one is better for writing larger scale more productized simulation programs In the next sections we examine both methods 6 19 1 Generating NED files Text processing programs like awk or perl are excellent tools to read in textual data files and generate NED files from them Perl also has extensions to access SQL databases so it can also be used if the network topology is stored in a database The advantage is that the necessary awk or perl program
278. o check whether t heNode s parameters and gates exist and are used correctly The node_type parameter will probably be given an input value somewhere higher in the module hierarchy so that the actual module type can be specified in the ini file or entered interactively The GeneralNode module type does not need to be implemented in C because no instance of it is created it is merely used to check the correctness of the NED file On the other hand the actual module type that will be substituted i e SwitchingNode in our case does not need to be declared in the NED files The like phrase enables the user to create families of modules that serve similar purposes and implement the same interface they have the same gates and parameters and to use them interchangeably in NED files This scheme directly parallels with the concept of polymorphism used in object oriented programming Submodule parameters Right after the declaration the values for the parameters of the declared submodules can be specified Example module ManyParameters parameters parl par2 switch submodules Submodl Node parameters pl 10 Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 41 p2 parlt par2 p3 switch 0 parl par2 Ld mean endmodule Expressions are mostly C style and they can contain parameters of the compound module being defined A separate section is dedicated to expressions He
279. o start working by itself without first receiving a message from other modules Programming with handleMessage To use the handleMessage mechanism in a simple module you must specify zero stack size for the module This is important because this tells OMNeT that you want to use handleMessage andnot activity Message event related functions you can use inhandleMessage e send family of functions to send messages to other modules e scheduleAt to schedule an event the module sends a message to itself e cancelEvent to delete an event scheduled with scheduleAt You cannot use the receive family and wait functions in handleMessage because they are coroutine based by nature as explained in the section about activity You also cannot use end because its job is to terminate the coroutine You have to add data members to the module class for every piece of information you want to preserve This information cannot be stored in local variables of handleMessage because they Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 75 are destroyed when the function returns Also they cannot be stored in static variables in the function or the class because they would be shared between all instances of the class Data members to be added to the module class will typically include things like e state e g IDLE BUSY CONN_DOWN CONN_ALIVE
280. odules you can hide the internal complexity of your model from direct inspection However nothing can prevent a user from building a simulation executable with it and exploring the structure of your compound module using OMNeT simulation kernel functions Consequently using precompiled compound modules is more useful as a structuring tool 5 8 Some simulation techniques 5 8 1 Modeling computer networks The hierarchical module structure of OMNeT allows you to organize the model around different levels Physical topology 1 Top level network 2 Subnetwork site 3 LAN 4 node Within a node 1 OSI layers The Data Link Network Transport Application layers are of greater importance 2 Applications protocols within a layer The advantage of OMNeT over many existing simulators is that the depth of the module nesting is not limited and what is in connection with the previous one that a simple module can be Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 90 transformed into a compound module by splitting the code into several simple modules without affecting existing users of the module and vica versa The latter means that the programmer of the model is not under pressure from possibly incorrect early design decisions about what to implement with a single module and what with a compound module 5 8 2 Modeling multiprocessor systems One can make use of flexible model topol
281. oe Sail 239 WSs 2 class Lambe DAS ter ator ite aynag vs cde a a a nasa woe 241 PB LU Br ASS CISA ft yc ges hank a a adie dla ls cate tle ton aah ile Sete lice 241 DDT Ae VAS NV AUCH ces ks wists ul Baise sic ds Sle dy loses EE a ceca 242 15 11 Output vector class cCOutVector y ccsciecsisissscaccadeasaascruesadiiasczesiacenandoataceuinsaedeaneoozanesens 243 15 111 class COUV CClOr siiictssiscca daasiter enian i n e E E EE 243 15 11 2 class COULPIEMET istinin p a a ias 244 15 12 Statistical data collection cStatistic and derived classes 0 eeecceeeseeeeeteeeeeteeeeneeees 245 Tl 2 wy ClASS C S AUIS fai GEN al en EUG E I II ie 246 15 122 las S26 SCADICY eena vances cueatnds denen E sus ucaae soegets ta dese ous E S 247 1512 3 Class CW ere ated Sb De visas aus a a a aay tae E ess ee aii 248 1S 12 AA classic Density HSL BASE anae a a aas ae 249 15 12 5 Class CHistOsrain Base 25 553 GA Sica ene SiG Be ES 250 V5 12 BC Tags Vir ISU STAN sine nnen egal Goss tedios dae sdoad E E N E EE SS 250 151257 Class CHGGHISEG Sram ASC is 5 ees yaes na a a r a i 250 1512 8 Class CliGng Histo anaes lt accsscgseceiat sa ceqssaaraseusdseceansas uae ia a 250 1512 9 Class cDoubleHist graM joss 5 ou sacs es as e ene a oaeaeei 251 15 13 The P and k split algorithms cPSquare CKSplit c cccccsssesssessessseessecstessessseeseesseen 252 15 131 el ss cPSgquare sst iite ern nita i e r a e ae a 252 15 13 2 elass CRSpliti rinena s a A ai a Eaa 253 15 14 Transie
282. ogies It is straightforward to create ring mesh butterfly torus hypercube tree fat tree and other topologies with conditional loop connections Furthermore general topology templates e g mesh or hypercube can be created where the types of the actual nodes are left as parameters The actual node types are substituted as parameter values for each concrete simulation Topology templates could be placed in a library and imported from there if needed 5 8 3 Parameter tuning Tuning means finding the parameter values which produce optimal operation of the system In OMNeT you can tune the model during runtime The code that monitors performance and changes parameter values can be placed e inside the model In this case the code does not necessarily form separate module s you can add the extra code to any already existing module e outside the model of the actual system If you choose this method you will create new modules that monitor and control the model OMNeT supports the model tuning concept by providing reference parameters Parameters that influence the model performance and need to be tuned will be declared at the highest layer and taken by both the model and the monitor part An example of model tuning is how one can determine the critical throughput of a communication network by changing the offered load according to performance measures of the network queuing times etc 5 8 4 Multiple experiments within one simu
283. ols like Token Ring and FDDI where the stations repeat the bits of a frame that arrives on the ring immediately without waiting for the whole frame to arrive in other words frames flow through the stations being delayed only a few bits If you want to model such networks the data rate modeling feature of OMNeT cannot be used If a message travels along a route through successive links and compound modules the model behaves as if each module waited until the last bit of the message arrives and only start its transmission then Technical University of Budapest Dept of Telecommunications OMNeT Manual Overview 22 tA tg tc sendin arrival Fig 2 4 Message sending over multiple channels Since the above effect is usually not the desired one typically you will want to assign data rate to only one connection in the route 2 1 5 Parameters Modules can have parameters Parameters are used for three purposes 1 to parameterize module topology 2 to customize simple module behaviour 3 for module communication as shared variables Parameters can take string numeric or pointer values numeric values include expressions using other parameters and calling C functions random variables from different distributions and values input interactively by the user Numeric valued parameters can be used to construct topologies in a flexible way Within a compound module parameters can define the number of submodules number of gates
284. om numbers e g the function takes mean and standard deviation and returns random variable of a certain distribution Reverse Polish expression Expression can contain constants cPar objects refer to other cPars e g module parameters can use many math operators etc function calls function must take 0 1 2 or 3 doubles and return a double The expression must be given is in an sXElem array see later random variable generated from a distribution collected by a statistical data collection object derived from cStatistic pointer to a non cObject item C struct non cObject object etc Memory management can be controlled through the configPointer member function Technical University of Budapest Dept of Telecommunications 126 OMNeT Manual The Simulation Library 127 object setObjectValue cObject pointer to an object derived from pointer cObject objectValue cObject op cObject op cObject Ownership management is done through takeOwnership indirect SetRedirection cPar value is redirected to another cPar value pod bse ed eee object All value setting and reading cPar redirection cancelRedirectionrt operates on the other cPar even the type function will return the type in the other cPar so you ll never get P as the type This redirection can only be broken with the cancelRedirection member function Module parameters taken by REF u
285. ompiler If a simulation program uses random numbers for more than one purpose the numbers should come from different random number generators OMNeT provides several independent random number generators by default 32 this number can be changed in defs h To avoid unwanted correlation it is also important that different simulation runs and different random number sources within one simulation run use non overlapping series of random numbers so the generators should be started with seeds well apart For selecting good seeds the seedtool program can be used it is documented later The random number generator was taken from JAIN91 pp 441 444 455 It has the following properties e Range 1 2 2 e Period length 2 12 e Method x x 7 mod 2 1 e To check if x 0 1 then x 10000 1 043 618 065 e Required hardware exactly 32 bit integer arithmetics The concrete implementation long intrand const long int a 16807 q 127773 r 2836 seed a seed q r seed q if s d lt 0 seed INTRAND_MAX return seed 6 9 1 Using random number generators directly The generator is directly accessible through the intrand function Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 117 long rnd intrand in the range 1 INTRAND_MAX 1 The random number seed can be specified in the ini file rcandom seed or set directly from withi
286. on Symposium ESS 98 October 26 28 1998 Nottingham UK Parameterized Topologies for Simulation Programs Andras Varga In the Proceedings of the Western Multiconference on Simulation WMC 98 Communication Networks and Distributed Systems CNDS 98 January 11 14 1998 San Diego CA The K Split Algorithm for the PDF Approximation of Multi Dimensional Empirical Distributions without Storing Observations Andras Varga and Babak Fakhamzadeh In Proceedings of the 9th European Simulation Symposium ESS 97 pp 94 98 October 19 22 1997 Passau Germany Flexible Topology Description Language for Simulation Programs Andras Varga and Gy rgy Pongor In Proceedings of the 9th European Simulation Symposium ESS 97 pp 225 229 October 19 22 1997 Passau Germany Technical University of Budapest Dept of Telecommunications OMNeT Manual References 258 Former OMNeT documents OMN1 Vass Zolt n PVM Extension of OMNeT to Support Statistical Synchronization Diploma Thesis Technical University of Budapest 1996 in Hungarian OMN2 Andr Maurits George van Montfort and Gerard van de Weerd OMNeT extensions and examples Technical University of Budapest Dept of Telecommunications 1995 OMN3 Jan Heijmans Alex Paalvast Robert van der Leij Network simulation using the JAR compiler for the OMNeT simulation system Technical University of Budapest Dept of Telecommunications 1995 OMN4 Varga Andr s
287. onnection that has an assigned bit error rate ber e The priority attribute is used by the simulation kernel to order messages in the message queue FES that have the same arrival time values e The time stamp attribute is not used by the simulation kernel you can use it for purposes like remembering the time when the message was enqueued or re sent e Other attributes and data members make simulation programming easier they will be discussed later parameter list encapsulated message context pointer e A number of read only attributes store information about the message s last sending scheduling source destination module and gate sending scheduling and arrival time They are mostly used by the simulation kernel while the message is in the FES but the information is still in the message object when a module receives the message Basic usage A cMessage object can be created in the following way cMessage msg new cMessage msg name kind length priority errorflag The kind Length and priority are integers and errorflag is boolean All arguments have default values so the following initializations are also valid cMessage msgl new cMessage Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 97 cMessage msg2 new cMessage data packet DATAPACKET_KIND 8 1500 Once a message has been created its data members can be changed by
288. or int j 0 j lt node gt outLinks J sTopoNode neighbour node gt out j gt remoteNode cGate gate node gt out j gt localGate ev lt lt lt lt neighbour gt module gt fullPath lt lt through gate lt lt gate gt fullName lt lt endl The nodes member function 1st line returns the number of nodes in the graph and node i returns a pointer to the ith node an sTopoNode structure The correspondence between a graph node and a module can be obtained by sTopoNode node topo nodeFor module cModule module node gt module The nodeFor member function returns a pointer to the graph node for a given module If the module is not in the graph it returns NULL nodeFor uses binary search within the cTopology object so it is fast enough sTopoNode s other member functions let you determine the connections of this node inLinks outLinks return the number of connections in i and out i return pointers to graph edge objects By calling member functions of the graph edge object you can determine the modules and gates involved The remoteNode function returns the other end of the connection and localGate remoteGate localGateId and remoteGateId return the gate pointers and ids of the gates involved Actually the implementation is a bit tricky here the same graph edge object sTopoLink is returned either as sTopoLinkIn or as sTopoLinkOut so t
289. or cPar The following operators are overloaded lt gt lt gt and These operators help using cPar objects in arithmetic expressions with the same syntax as internal numeric types Two different sets of overloaded arithmetic operators are provided mutually excluding the other One works through converting the cPar to double and the other through converting to long The double version is the default but you can choose the other set by defining LONG_CPAR_OPERATIONS before including cpar h If neither set is needed define NO_CPAR_OPERATIONS 15 8 3 struct sXElem If the value of the cPar is of expression type the expression must be converted to reversed Polish form The reversed Polish form expression is stored in a vector of sXElem structures sXElem is not a descendant of cObject void operator int _i void operator long _1 void operator double _d Effect during evaluation of the expression pushes the given number which is converted to double on the evaluation stack void operator cPar _v Effect during evaluation of the expression takes the value of the cPar object a double and pushes the value on the evaluation stack The cPar is an external one its ownership does not change This is how NED language REF parameters in expressions are handled void operator cPar amp _r Effect during evaluation of the expression takes the value of the cPar object a double and pushes the value o
290. or other models too e propagation delay sec e bit error rate errors bit e data rate bits sec Each of these parameters is optional One can specify link parameters individually for each connection or define link types also called channel types once and use them throughout the whole model Technical University of Budapest Dept of Telecommunications OMNeT Manual Overview 21 The propagation delay is the amount of time the arrival of the message is delayed by when it travels through the channel Propagation delay is specified in seconds The bit error rate has influence on the transmission of messages through the channel The bit error rate is the probability that a bit is incorrectly transmitted Thus the probability that a message of n bits length is transferred correctly is P sent message received properly 1 ber where ber bit error rate and n number of bits in message The message has an error flag which is set in case of transmission errors The data rate is specified in bits second and it is used for transmission delay calculation The sending time of the message normally corresponds to the transmission of the first bit and the arrival time of the message corresponds to the reception of the last bit A B tA tg message sending transmission delay propagation delay Bal AE EEEE EE message arrival Fig 2 3 Message transmission The above model is not applicable for modeling some protoc
291. os occur in the code generated by the NEDC compiler one for each channel definition cHead This is only dummy object it stands for the current locals module s local object list any object cHead List of all other lists superhead cHead Technical University of Budapest Dept of Telecommunications OMNeT Manual The Design of OMNeT 183 11 3 4 The coroutine package The coroutine package is in fact two coroutine packages There is a platform independent coroutine package that creates all coroutine stacks inside the main stack It was taken from KOF85 It allocates stack by deep deep recursions and then plays with setjmp and longjmp to switch from one another Its drawback is that under 16 bit Intel platforms DOS real mode and Win16 stack is limited to 64K which is not very much The other package allocates stack by malloc and uses a short assembly code to initialize it for the first use Then it also uses set jmp and long jmp This is implemented under DOS BC3 1 and also RISC6000 where the original set jmp long jmp behaved in an unfriendly way and the portable coroutine package could not be used The coroutines are represented by the cCoroutine class cSimpleModule has cCoroutine as one a base class 11 3 5 Object ownership contains relationships Ownership Exclusive right and duty to delete the child objects Ownership works through cObj s ownerp prevp nextp and firstchildp l
292. oth in OMNeT and OPNET Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix A OPNET and OMNeT 187 Model hierarchy levels network level subnetwork nesting arbitrary levels of submodule nesting possible node level no nesting process level no nesting Topology description method OPNET provides two tools for defining module topology graphical editors to design network and node level models and EMA External Model Access an API for building model files from C programs These tools correspond to OMNeT s tools in the following way Graphical graphical editor within the graphical editor GNED IDE Low level EMA C output of NED compilation There is no high level textual model description in OPNET like NED is in OMNeT This means that one has either to use the graphical editor or write lengthy C code using the EMA API The OPNET graphical model editor can only create fixed non parameterized topologies There s a significant difference between how EMA and OMNeT s NED are used OPNET s EMA generates model files EMA applications are standalone programs one writes the EMA C code compiles and runs it and the EMA executable will generate a model file that can be read into the graphical editor or loaded by simulation programs EMA cannot be used from within a simulation program In contrast the compiled NED code of OMNeT becomes part of the simulation program
293. ould become negative an error occurs The encapsulatedMsg function returns a pointer to the encapsulated message or NULL if no message was encapsulated 6 3 4 Information about the last sending Readonly attributes The following functions exist in cMessage bool isSelfMessage cGate senderGate return NULL if scheduled cGate arrivalGate or unsent message int senderModuleld int senderGatelId int arrivalModuleId int arrivalGatelId simtime_t creationTime Simtime_t sendingTime simtime_t arrivalTime bool arrivedOn int g bool arrivedOn const char s int g 0 TBD comments Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 100 Context pointer cMessage contains a void pointer which is set returned by the setContextPointer and contextPointer functions void context msg gt setContextPointer context void context2 msg gt contextPointer It can be used for any purpose by the simulation programmer It is not used by the simulation kernel and it is treated as a mere pointer no memory management is done on it Intended purpose a module which schedules several self messages timers will need to identify a self message when it arrives back to the module ie the module will have to determine which timer went off and what to do then The context pointer can be made to poi
294. out for me under Linux to get a shell with a 64M stack limit su root Password ulimit s 65536 su andras ulimit s 65536 If you do not want to go through the above process at each login you can change the limit in the PAM configuration files In Redhat Linux maybe other systems too add the following line to etc pam d login session required lib security pam_limits so and the following line to etc security limits conf hard stack 65536 A more drastic solution is to recompile the kernel with a larger stack limit Edit usr src linux include linux sched h and increase _STK_LIM from 8 1024 1024 to 64 1024 1024 Finally it youre tight with memory you can switch to Cmdenv Tkenv increases the stack size of each module by about 32K so that user interface code that is called from a simple module s context can be safely executed Cmdenv does not need that much extra stack 8 6 2 Memory allocation problems For investigating memory allocation problems try using Cmdenv and uncomment the defines in src envir cmdenv heap cc HEAPCHECK checks heap on new delete COUNTBLOCKS counts blocks on heap and tells it if none left ALLOCTABLE remembers pointers and reports heap contents if only LASTN blocks remained DISPLAYALL reports every new delete DISPSTRAYS reports deleting of pointers that were not registered by operator new or that were deleted since then BKPT calls a function at a specified new del
295. pest Dept of Telecommunications OMNeT Manual Simple Modules 67 As an example the class declaration class SlidingWindow public cSimpleModule Module_Class_Members SlidingWindow cSimpleModule 8192 l expands to something like this class SlidingWindow public cSimpleModule public SlidingWindow const char name cModule parentmodule unsigned stacksize 8192 cSimpleModule name parentmodule stacksize virtual const char className return SlidingWindow Expanded form of the constructor You will implement e aconstructor with the argument list const char name cModule parentmodule unsigned stacksize stacksize e aclassName function which returns the name of the class as char The advantage is that you get full control over the constructor so you can initialize data members of the class if you have any You should not change the number or types of the arguments taken by the constructor because it is called by OMNeT generated code Also remember to overwrite the className function An example class TokenRing_MAC public cSimpleModule public cQueue queue a data member TokenRing_MAC const char name cModule parentmodule unsigned stacksize 8192 virtual const char className return TokenRing_MAC TokenRing_MAC const char name cModule parentmodule unsigned stacksize cSimpleModule name parentmodule stacksize que
296. pos ypos at most itemsperrow submodules in a row keeping deltax and deltay distances Defaults itemsperrow 5 deltax deltay are chosen so that submodules do not overlap Used for module vectors Arranges submodules in an ellipse with the top left corner of its bounding boxes at xpos ypos with the width and height Defaults width 40 height 24 Rectangle with the given height and width Defaults width 40 height 24 Ellipse with the given height and width Defaults width 40 height 24 Specifies options for the rectangle or oval Any valid Tk color specification is accepted English color names or rgb rrggbb format where r g b are hex digits Defaults fillcolor 8080ff a lightblue outlinecolor black borderwidth 2 Use the named icon No default If no icon name is present box is used Tags used in enclosing module display strings Tag p xpos ypos b width height rect Meaning Place enclosing module at xpos ypos pixel position with 0 0 being the top left corner of the window Display enclosing module as a rectangle with the given height and width Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language b width height oval o fillcolor outlinecolor borderwidth Tags used in connection display strings Tag m auto m north m west m east m south 58 Defaults width height are chosen automatically Display enclosi
297. pper bounds This will be interpreted as nested for statements the leftmost index being the outermost and the rightmost index being the innermost loop for i 0 4 j 0 4 do Lees endfor One can also use an index in the lower and upper bound expressions of the subsequent indices for i 0 3 j i 1 4 do Elwes endfor In the above example the following i j pairs will be used for the connections inside the for statement 0 1 0 2 0 3 0 4 1 2 1 3 1 4 2 3 2 4 3 4 A gate cannot be used in more than one connection and one connection cannot be made more than once Consider the following bogus statement for i 0 2 0 2 do modulel out i gt module2 in 1 endfor It will cause a runtime error each connection is made twice as the index variable j is not used in the connection In general every connection inside a loop should use all the index variables at both sides of the connection Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 45 Conditional connections Connections can be conditional This is a conditional connection for i 0 n do Sender outgate i gt Receiver i ingate if 1 2 0 endfor This way we connected every second gate The nocheck modifier Conditional connections are especially useful with random numbers when they can create random connections Here a problem can be that by default the simulation program checks if al
298. process This process will be a multithreaded process with each thread running on a physically different processor of the hardware Each thread will be responsible for simulating one segment of the model different threads communicate via shared memory logically with PDES protocols This practically means that you can run Parsec parallel simulation on shared memory multiprocessors e Quite differently OMNeT supports distributed simulation where the simulation program consists of separate operating system processes typically running on different hosts interconnected via a LAN These processes communicate by passing messages over the LAN by using the PVM library This difference accounts for the usage differences While Parsec s parallel simulation support is almost transparent to the user OMNeT s distributed simulations are inherently more difficult to set up and manage you have to install PVM etc Parsec supports conservative PDES OMNeT supports conservative PDES and Statistical Synchronization 13 4 Feature summary Feature OMNeT PARSEC Programs components graphical model editor GNED result analysis plotting Plove interactive execution Tkenv tracing parameter file omnetpp ini random numbers support seedtool Model structure encapsulation grouping compound modules Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix B Parsec and OMNeT connec
299. ption is especially convenient is when the process has many states but transitions are very limited ie from any state the process can only go to one or two other states For example this is the case when programming a network application which uses a single network connection The pseudocode of the application which talks to a transport layer protocol might look like this activity while true open connection by sending OPEN command to transport layer receive reply from transport layer if open not successful wait some time continue loop back to while while there s more to do send data on network connection if connection broken continue outer loop loop back to outer while wait some time receive data on network connection if connection broken continue outer loop loop back to outer while wait some time close connection by sending CLOSE command to transport layer if close not successful handle error wait some time If you want to handle several connections simultaneously you may dynamically create as instances of the simple module above as needed Dynamic module creation will be discussed later Activity is run as a coroutine Activity is run in a coroutine Coroutines are a sort of threads which are scheduled non preemptively this is also called cooperative multitasking From one coroutine you can switch to Technical University of Budapest Dept of Te
300. r connection arrow at run time by calling setDisplayString The display string of a connection arrow is stored in its source gate Display string changes will immediately take effect Examples setDisplayString dispSUBMOD p 100 100 b 60 30 rect o red black 3 parentModule gt setDisplayString dispSUBMOD p 100 yy gate out gt setDisplayString o yellow 3 TBD immediate arg to setDisplayString cMessage s setDisplayString 6 18 Tips for speeding up the simulation Here are a few tips that can help you make the simulation faster e Turn off the display of screen messages when you run the simulation You can do this in the ini file Alternatively you can place ifdefs around your ev lt lt and ev printf calls and turn off the define when compiling the simulation for speed Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 144 e Store the module parameters in local variables to avoid calling cPar member functions every time e Use gate numbers instead of gate names e Try to minimize message creations and deletions Reuse messages if possible e Do not give name strings to objects that are created and deleted many times pass NULL pointer as name e Use numeric index to get an object from a cArray not the object name You can do this also with message parameters Two techniques are discussed here in detail e messag
301. ractive simulation Tracing and debugging powerful command line separate window for each debugger ODB module s output single steps run until inspectors snapshot etc Tkenv Animation mostly used in record interactive execution with playback mode message flow animation statistics animation etc animation spec must be P Tkenv given in advance via anim probes Interactive simulation not supported strongly supported via object inspectors and watches Tkenv Random numbers ee OMNeT o O Distributions EEEE many a in distributions Sa built in distributions through algorithms as C functions Additional distributions through histograms as C functions algorithms or through histograms Random number one random number several independent generation generator no support for random number generators seed selection tool to support selecting good seed values OPNET has many built in distributions implemented with algorithms C functions Additional distributions are supported as histograms There is only one common source of random numbers OPNET has no aid for selecting seed values that produce long non overlapping random number sequences OMNeT only four basic distributions are provided They are implemented as C functions Additional distributions can be added by the user and they are treated exactly in the same way as built in ones Defining and using distributions in histogram form is a
302. rating over submodules To access all modules within a compound module use cSubModIterator For example for cSubModIterator submod parentModule submod end submod ev lt lt submod gt fullName submod is pointer to the current module the iterator is at The above method can also be used to iterate along a module vector since the name function returns the same for all modules for cSubModIiterator submod parentModule submod end submod t if submod gt isName name if submod is in the same vector as this module int its _index submod gt index do something to it Walking along links To determine the module at the other end of a connection use cGate s fromGate toGate and ownerModule functions For example cModule neighbour gate outputgate gt toGate gt ownerModule For input gates you would use fromGate instead of toGate 6 7 Dynamic module creation Why If you do not know how many modules you ll need you can create modules dynamically and dispose of them when they are no longer needed Both simple and compound modules can be created this way If you create a compound module dynamically all its submodules will be recursively built Let s suppose you are implementing a transport protocol for a computer network model It is convenient to have a separate module to handle each connection However there s
303. re only the modes of parameter passing are discussed The default parameter passing method is by value However the user can write ref or ancestor before the parameter name Writing ref means that the parameter is not passed by value but by reference This means that instead of the value of the parameter the address of the parameter is passed Writing ancestor before the parameter name means that the parameter will be searched upwards among the parameters of all future enclosing modules of the current module This reference cannot be resolved or checked by the NEDC compiler it can only be done at runtime when the whole network has been built up The parameter which is found first is used if no such parameter can be found in any of the enclosing modules the system will give an error during runtime The ancestor and ref modifiers are independent they can be used together For example simple sub_sub parameters s_s_parl s_s_par2 endmodule sub_sub module sub parameters S_par submodules child sub_sub parameters s_s_parl ref s_par S_S_par2 ref ancestor m_par2 endmodule sub module mod parameters m_parl m_par2 submodules child sub parameters S_par m parl endmodule mod Again note that the network description compiler can check for the existence of ordinary parameters but not for ancestor parameters it cannot predict in what modules the current module will be embedded in an actual network descriptio
304. re are many possible subsequent states for any state Such algorithms are difficult to program with activity or the result is code which is better suited for handleMessage see rule of thumb below Most communication protocols are like this There s also a good rule of thumb If your module programmed withact ivity looks like this activity initialization code while true msg receive arbitrary code which doesn t contain any receive or wait calls Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 76 Then it can be trivially converted to handleMessage initialize initialization code handleMessage msg arbitrary code which doesn t contain any receive or wait calls Example 1 Simple traffic generators and sinks The code for simple packet generators and sinks programmed with handleMessage might be as simple as this PacketGenerator handleMessage m create and send out packet schedule m again to trigger next call to handleMessage self message PacketSink handleMessage m delete m Note that PacketGenerator will need to redefine initialize to create m and schedule the first event The following simple module generates packets with exponential inter arrival time Some details in the source haven t been discussed yet but the code is probably understandable nevertheless class Generator public cS
305. ridden at run time by the Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 168 OMNETPP_TKENV_DIR environment variable The existence of a separate script library can be inconvenient if you want to carry standalone simulation executables to different machines To solve the problem there is a possibility to compile the script parts into Tkenv as a large string constant The details the tcl2c program its C source is there in the Tkenv directory is used to translate the tcl files into C code tclcode cc which gets included into tkapp cc On Unix this feature is enabled in Tkenv s makefile it is documented there exactly how On Win95 NT one has to manually compile tcl2c c into tcl2c exe run it to produce tclcode cc and then compile tkapp cc without providing the OMNETPP_TKENV_DIR external define The latter will cause tkapp cc to include and use tclcode cc 8 5 3 In Memoriam There used to be other user interfaces which have been removed from the distribution e TVEny A Turbo Vision based user interface the first interactive UI for OMNeT Turbo Vision was an excellent character graphical windowing environment originally shipped with Borland C 3 1 e XEnv A GUI written in pure X Motif It was an experiment written before I stumbled into Tcl Tk and discovered its immense productivity in GUI building XEnv never got too far because it was really very
306. ring the libraries If you need to recompile the OMNeT libraries with different flags e g for debugging or optimized for speed then cd to the top level omnetpp directory edit configuser vc accordingly then say nmake f Makefile vc clean nmake f Makefile vc If you want to recompile just a single library then cd to the directory of the library and type nmake f Makefile vc clean nmake f Makefile vc 7 4 6 Using Plove If you want to use Plove you should download and install Gnuplot You ll also need a couple of Unix tools like grep and awk the easiest way to get them is to download and install the Cygwin package from www cygnus com When you have everything installed start Plove and set the appropriate configuration in OptionslExternal programs If you entered everything correctly Plove should work A usual caveat is that Gnuplot expects forward slashes in filenames and Plove supplies backslashes or vica versa there are multiple incompatible builds of Gnuplot on NT if you suspect this might be the problem reverse the slash backslash setting in OptionslExternal programs Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 155 8 RUNNING THE SIMULATION 8 1 Command line switches An OMNeT executable accepts the following command line switches h The program prints a short help message and the networks contained in the executable and exits f lt fileNa
307. rm associates the class with a NED simple module declaration of a different name You can use this form when you have several modules which share the same interface This feature will be discussed in detail in the next section Header files Module declarations should not be put into header files because they are macros expanding to lines for which the compiler generates code Compound modules All module types including compound modules need to have module declarations For all compound modules the NEDC compiler generates the Define _Module lines automatically However it is your responsibility to put Define _Module lines into one of the C sources for all your simple module types Implementation Unless you are dying to learn about the dirty internals you may just as well skip this section But if youre interested here it is Define_Module and also Define_Module_Like is a macro which expands to a function definition plus the definition of a global object something like this ugly code luckily you won t ever need to be interested in it static cModule MyClass__create const char name cModule parentmod return cModule new MyClass name parentmod cModuleType MyClass__type MyClass MyClass ModuleCreateFunc MyClass__create The cModuleType object can act as a factory it is able to create an instance of the given module type This together with the fact that all cModuleType objects are
308. s cMessage and cPacket Header file cmessage h cpacket h sim directory cMessage is the message class in OMNeT cMessage can be assigned a name a property inherited from cOb ject and it has other attributes including message kind length priority error flag and time stamp After being sent through a channel cMessage also remembers the sending and delivery times and its source module cMessage holds a cArray which means that a cMessage can be attached any number of objects These objects will typically be of cP ar type but other types are also possible 15 6 1 Message base class cMessage Construction destruction copying cMessage cMessage amp msg Copy constructor creates an exact copy of the argumentmsg cMessage const char name NULL int k 0 int len 1 int pri 0 bool err false Constructor It accepts name message kind length priority and error flag as arguments virtual cObject dup cMessage amp operator cMessage amp msg Duplication and the assignment operator work all right with cMessage Redefined virtual functions virtual const char className Returns pointer to the class name string cMessage Message properties void setKind int k Sets message kind The message kind member is not used by OMNeT it can be used freely by the user void setPriority int p Sets message priority The priority member is used when the simulator inserts messages in the message queue
309. s message generator sink concentrator simple hub queue etc e Modules that implement routing algorithms in a multiprocessor or network To create a library you compile the simple module C sources and collect the object files in one directory You ll also need to the provide the NED descriptions library generator ned generator o sink ned sink o ethernet ned ethernet o The NED files contain the interfaces of the simple modules For example generator ned Simple Generator parameters interarrival time message_length message_kind gates out output endsimple The user of the library would include generator ned in his her NED files and link the executable with generator o This is more or less the same concept as conventional C C header files and libraries The basic advantage is the same as with C C you save compilation time and hide concrete implementation The latter also means that you can give the module library to others without having to share the C source Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 88 It could also be meaningful to provide the C header files with the module class declarations This would enable the user to directly call the member functions of the module object from the simulation program and derive new module classes by redefining the virtual functions You do not need to have a separate NED file for each module you could merg
310. s an ancient game with two players and a bunch of sticks The players take turns removing 1 2 3 or 4 sticks from the heap of sticks at each turn The one who takes the last stick is the loser Of course building a model of the Nim game is not much of a simulation project but it nicely demonstrates the modeling approach used by OMNeT Describing the model consists of two phases e topology description e defining the operation of components 3 1 Topology The game can be modelled in OMNeT as a network with three modules the game a manager module and two players The modules will communicate by exchanging messages The game module keeps the current number of tokens and organizes the game in each turn the player modules receives the number of tokens from the Game module and sends back its move Fig 3 1 Module structure for the Nim game Playerl Player2 and Game are simple modules e g they have no submodules Each module is an instance of a module type We ll need a module type to represent the Game module let s call it Game too We can implement two kinds of players SnartPlayer which knows the winning algorithm and SimplePlayer which simply takes a random number of sticks In our example Player1 will be aSmartPlayer and Player2 willbeaSimplePlayer The enclosing module Nim is a compound module it has submodules It is also defined as a module type of which one instance is created the system module
311. s are at the same level there s no submodule nesting Connections which span across compound module boundaries are also represented as one graph edge Graph edges are directed just as module gates are If youre writing a router or switch model the cTopology graph can help you determine what nodes are available through which gate and also to find optimal routes The cTopology object can calculate shortest paths between nodes for you The mapping between the graph nodes edges and network model modules gates connections is preserved you can easily find the corresponding module for acTopology node and vica versa 6 8 2 Basic usage You can extract the network topology into a cTopology object by a single function call You have several ways to select which modules you want to include in the topology e by module type e by a parameter s presence and its value e with a user supplied boolean function First you can specify which node types you want to include The following code extracts all modules of type Router or User Router and User can be both simple and compound module types cTopology topo topo extractByModuleType Router User NULL Any number of module types up to 32 can be supplied the list must be terminated by NULL Second you can extract all modules which have a certain parameter topo extractByParameter ip_address You can also specify that the parameter must have a certain value for th
312. s of mobile entities Messages can contain arbitrarily complex data structures Simple modules can send messages either directly to their destination or along a predefined path through gates and connections The local simulation time of a module advances when the module receives a message The message can arrive from another module or from the same module self messages are used to implement timers Gates are the input and output interfaces of modules messages are sent out through output gates and arrive through input gates Each connection also called link is created within a single level of the module hierarchy within a compound module one can connect the corresponding gates of two submodules or a gate of one submodule and a gate of the compound module Fig 2 2 parent module parent module 1 s2 Submodules connected to each other Submodules connected to the parent module Fig 2 2 Connections Due to the hierarchical structure of the model messages typically travel through a series of connections to start and arrive in simple modules Such series of connections that go from simple module to simple module are called routes Compound modules act as cardboard boxes in the model transparently relaying messages between their inside and their outside world 2 1 4 Link characteristics Connections can be assigned three parameters which facilitate the modeling of communication networks but can be useful f
313. s patterns the first matching occurrence is used An example ini file omnetpp ini Parameters token num_stations 3 token num_messages 10000 Run 1 token stations wait_time 10ms Run 2 token stations 0 wait_time 5ms token stations wait_time 1000ms 8 2 4 Configuring output vectors As a simulation program is evolving it is becoming capable of collecting more and more statistics The size of output vector files can easily reach a magnitude of several ten or hundred megabytes but very often only some of the recorded statistics are interesting to the analyst In OMNeT you can control how cOut Vector objects record data to disk You can turn output vectors on off or you can assign a result collection interval Output vector configuration is given in the OutVectors section of the ini file or in the Run 1 Run 2 etc sections individually for each run By default all output vectors are turned on Entries configuring output vectors can be like that module pathname objectname enabled yes no module pathname objectname interval start stop module pathname objectname interval stop module pathname objectname interval start The object name is the string passed to cOut Vector in its constructor or with the setName member function cOutVector eed End to End Delay 1 Start and stop values can be any time specification accepted in
314. s pointer to the object s name The function never returns NULL rather it returns ptr to nn virtual const char fullName To be redefined in descendants E g see cModule fullName virtual const char fullPath int 1 Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 216 Returns the name of the object with its place in the object ownership hierarchy net comp modem 5 baud rate See also cModule fullPath bool isName const char s Returns t rue if the object s name is identical with the string passed User interface information functions virtual const char className j Returns a pointer to the class name string CObject In derived classes usual implementation is return classname virtual void info char buf Copies a short description of the object into buf This function is used by the graphical user interface TkEnv See also Functions supporting snapshots virtual const char inspectorFactoryName Returns the name of the class which can create inspector windows for objects of this class e g in Tkenv Ownership control virtual cObject defaultOwner This function should return a pointer to the default owner of the object The function is used by the drop member function redefined in cOb ject descendant container classes cObject owner Returns pointer to the owner of the object void se
315. sage Parsec and C SIM use threads OMNeT supports both programming models the author does not know of another simulation tool that does so Debugging and Tracing Support What debugging or tracing facilities does the simulation tool offer Simulation programs are infamous for long debugging periods C based simulation tools rarely offer much more than printf style debugging often the simulation kernel is also capable of Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 16 dumping selected debug information on the standard output Animation is also often supported either off line record amp playback or in some client server architecture where the simulation program is the server and it can be viewed using the client Off line animation naturally lacks interactivity and is therefore little use in debugging The client server solution typically has limited power because the simulation and the viewer run as independent operating system processes and the viewer has limited access to the simulation program s internals and or it does not have enough control over the course of simulation execution OPNET has a very good support for command line debugging and provides both off line and client server style animation NetSim and Ptolemy use the client server method of animation OMNeT goes a different way by linking the GUI library with the debugging tracing capability into the simulation exec
316. se this mechanism 6 13 Statistics and distribution estimation 6 13 1 cStatistic and descendants There are several statistic and result collection classes cStdDev cWeightedStdDev Long Histogram cDoubleHistogram cVarHistogram cPSquare and cKSp1it They are all derived from the abstract base class cStatistic cStdDev keeps number of samples mean standard deviation minimum and maximum value etc cWeightedStdDev is similar to cStdDev but accepts weighted observations cWeightedStdDev can be used for example to calculate time average It is the only weighted statistics class cLongHistogram and cDoubleHistogram are descendants of cSt dDev and also keep an approximation of the distribution of the observations using equidistant equal sized cell histograms cVarHistogram implements a histogram where cells do not need to be the same size You can manually add the cell bin boundaries or alternatively automatically have a partitioning created where each bin has the same number of observations or as close to that as possible cPSquare is a class that uses the P algorithm described in JCH85 The algorithm calculates quantiles without storing the observations one can also think of it as a histogram with equiprobable cells cKSplit uses a novel experimental method based on an adaptive histogram like algorithm Published papers about k split can be downloaded from the OMNeT Web sit
317. sis or spreadsheet programs which provide numerous ways to display data as diagrams do calculations on them etc One could use for example Gnuplot Matlab Excel etc 9 3 3 Visualization under Unix Two programs are in common use Gnuplot and Xmgr Both are free and both have their good and bad sides will briefly discuss them There are innumerable tutorials and documentation about them on the Web some of them you will find among the References Both programs can eat files produced by splitvec Both programs can produce output in various forms on screen in Postscript format printer files Latex output etc For DTP purposes Postscript seems to be the most appropriate On Windows the easiest way is to copy the picture to the clipboard from the Gnuplot window s system menu Gnuplot has an interactive command interface To get the vectors in mysiml vec and mysim4 vec plotted in the same graph you can type plot mysiml vec with lines mysim4 vec with lines To adjust the y range you would type set yrange 0 1 2 replot There are several commands to adjust ranges plotting style labels scaling etc Gnuplot can also plot 3D graphs Gnuplot is also available for DOS Windows and other platforms Gnuplot also has a simple graphical interactive user interface called PlotMTV However we recommend that you use OMNeT s Plove tool described in an earlier section Xmer is an X Motif based program with a menu driven grap
318. sity of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 253 15 13 2 class cKSplit Implements the k split algorithm 15 14 Transient and result accuracy detection Header file cdetect h sim directory Detection of the end of the transient period and a certain result accuracy is supported by OMNeT The user can attach transient detection and result accuracy objects to a result object cStatistic s descendants 15 14 1 class cTransientDetection Virtual base class for transient detection 15 14 2 class cTDExpandingWindows An algorithm for transient detection Uses sliding window approach with two windows and checks the difference of the two averages to see if the transient period is over 15 14 3 class cAccuracyDetection Virtual base class for result accuracy detection 15 14 4 class cADByStddev An algorithm for result accuracy detection The actual algorithm divide the standard deviation by the square of the number of values and check if this is small enough 15 15 Finite State Machine cFSM cFSM is not yet documented here See the User Manual 15 16 Routing support cTopology cTopology is not yet documented here See the User Manual 15 17 The user interface cEnvir and ev Header file cenvir h envir directory cEnvir is the user interface class cEnvir is not a descendant of cObject and there is only one instance of cEnvir ev ev is a Static object All I O
319. sseeesssees 169 Oe ARO CULION Speed er ia aE E E E E EA E R NE 169 9 ANALYZING SIMULATION RESULTS ssessesessossesossosoesossosossossosoesossosossossosoesoseososseseoses 171 9 1 Plotting output vectors with Plove sssesseessesssesesseessseesseesseseseeessstesseessersseessseeessresseesse 171 91 T Plove feature Sinna n a E a E E E ca cel Si eh tata 171 Nea EE E a a a EA SE A 171 AL NV TA LIAL ELMS Doe eects cn Gh a a T R a E E A E A AN 172 9 2 Format of OUtput vector FES 5 6 sisisi ises i ino ae as iae iaia 172 93 Working without POV E a ee Ge SEE OGG BEE EEES 173 9 3 1 Extracting vectors from the TE 5 lt 0i52 4 lt 5 lt dsacctvsiesccanscee eat ta saceaeatoeeedecideseseiyaeeees 173 Fn WESTIN SPVIEV CO oye sey nae E E cu ont E E E eee cos eee ases 173 9 3 3 Vis a lizati n nder UNIX iernii eE EEEE Sa 174 10 PARALLEL EXECUTION csesscssrssisssistrssesarssoisssotesessorseos osoro rarse rariss 175 10 1 S ppott for parallel execution nsee a a aR Aa A A ER ARE scat esas 175 DY A TV VC Wc carl n n a E ncaa so o Sue A dawesen eet 175 KOREEN la OIE S 5 EEEE A E E EA E E E eccd ee acasaweases 175 10 1 3 The PVM virtual machine sssseessensseeeseeeesseesseesserssereseeessseesseesseesseeesseesssees 176 10 1 4 Mapping logical machines to physical ones ssessseeeseesesseessesssessseersseesssees 177 10 1 5 Configuration of the slaves ssnssenssesssseeesseessressesssereseeessseesseessersseeesseesssees 177 10
320. sses One of the important distinguishing features of PARSEC is its ability to execute a discrete event simulation model using several different asynchronous parallel simulation protocols on a variety of parallel architectures Thus with few modifications a PARSEC program may be executed using the traditional sequential Global Event List simulation protocol or one of many parallel protocols In reality PARSEC currently supports only conservative PDES but not any optimistic protocol Andras In addition PARSEC provides powerful message receiving constructs that result in shorter and more natural simulation programs The PARSEC language has been derived from the Maisie language but with several improvements both in the syntax of the language and in its execution environment The PARSEC web site is at http pcl cs ucla edu It is very impressive 13 2 What is inside the Parsec package When you download and install the PARSEC distribution basically you find e pce the Parsec compiler and e 2 variants of the Parsec runtime library This shows that PARSEC is strictly a simulation and parallel programming language which is restricted to the area of entities messages and the tasks centered around message sending and receiving It is difficult to compare to OMNeT which is more of a complete simulation environment In OMNeT the simulation library alone covers much more functionality than the whole PARSEC PARS
321. ssesssesesseetsstesserssessseressteesseesstesseesseeesseeessresseesse 99 6 3 4 Information about the last sending 0 eee eeeececeeeeeceeeeeceeeeeceteeecneeeeneeeenaeeeeaas 99 6 3 5 The cPacket Class a vectivicstis s eileen tesia taa e ap isie 100 6 3 6 Subclassing cMessage and cPacket snnesessseeeseeeesseesseesseesseresseeessresseesseesseee 100 GA Sending and receiving messages e sssessssssessseressettssttssersseresteetsstessressersseeessessseesseesse 101 6 4 1 Sending MISS AGES ca cath xe ae a coed E N R A E sane te 101 GAZ DeLay ed Send A a ae e EAA A ETEA 101 6 4 3 Direct message Sen vias xs scares suartsat as cesses axaceusduaceantenscataetonse tea seqeusaiaaccaenminees 102 GAA RECEIVING INESSAC CS eG I soda vec pant sa E E ooh toate vedsas 102 6 4 5 The wait fUnCtion ceccccessccccccceceessssececeeececeesessnseceeeecesesesesseaeceeeeeeseeensaaees 103 GAG Sel eMC SS aes 55 soi cs ys ese E saves cay oon eases nso ec sin cnatee av aoe ens eco eases 104 6 4 7 Querying the state of an output Gate ee eeecssececeeececeteeeseteeeeseceeteeeetaeeees 105 6 4 8 Stopping the simula lon 54 555 GORGE ERR Sais 105 6 5 Accessing module parameters and Gates cccsescecsscecssccecsseceesscceesscesesscesenseeeeneeeees 106 6 3 Mod le pararietetS se rs phos Gace Beans 59569 Sara soca aere Beas Pasa cued geca somes gede ie ghes AE S 106 6 5 2 Gates and links irain a E EEE aa AN a ees 106 6 6 Walking the mo
322. st Dept of Telecommunications OMNeT Manual An Example The NIM Game 32 player_to_move 3 player_to_move while num_sticks gt 0 j ev lt lt nPlayer lt lt player_to_move lt lt lt lt player player_to_move lt lt won n endSimulation 3 3 Running the simulation Once the source files are ready one needs to compile and link them into a simulation executable One can specify the user interface to be linked Before running the simulation one can put parameter values and all sorts of other settings into an initialization file that will be read when the simulation program starts i file omnetpp ini 1 General network nim random seed 3 ini warnings false Cmdenv module messages yes verbose simulation no Suppose we link the NIM simulation with the command line user interface We get the executable nim nim exe under Windows When we run it we ll get the following screen output nim Or C OMNETPP SAMPLES NIM gt nim exe OMNeT Discrete Simulation TUB Dept of Telecommunications 1990 97 Preparing for Run 1 Setting up network nim Running simulation Let the game begin Player 1 SmartPlayer Player 2 SimplePlayer Sticks left 29 Player 2 SimplePlayer to move SimplePlayer is taking 2 out of 29 sticks Player 2 SimplePlayer took 2 stick s Sticks left 27
323. stopped at each breakpoint call in the simple modules update freq fast 10 update freq express 500 animation delay 0 3s Delay between steps when you slow execute the simulation Animation enabled yes animation msgnames yes animation msgcolors yes animation speed 1 0 Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 163 Slaves write slavelog yes slavelog file slave log module messages yes errmsgs to console yes infomsgs to console no modmsgs to console no 8 3 Choosing good seed values the seedtool utility For selecting good seeds the seedtool program can be used it is in the utils directory When started without command line arguments the program prints out the following help seedtool part of the OMNeT Simulation System BME HIT 1997 A tool to help select good random number generator seed values Usage seedtool i seed index of seed in cycle seedtool s index seed at index index in cycle seedtool d seedl seed2 distance of seedl and seed2 in cycle seedtool g seedO dist generate seed dist away from seed0 seedtool g seedO dist n generate n seeds dist apart starting at seed0d seedtool t generate hashtable seedtool p print out hashtable The last two options p and t were used internally to generate a hash table of pre computed
324. structor because OMNeT keeps track of objects you create and disposes of them automatically sort of automatic garbage collection However it cannot track objects not derived from cObject see later so they may need to be deleted manually from the destructor Multi stage initialization In simulation models when one stage initialization provided by initialize is not sufficient one can use multi stage initialization Modules have two functions which can be redefined by the user void initialize int stage int numInitStages At the beginning of the simulation initialize 0 is called for all modules then initialize l1 initialize 2 etc For each module numInitStages must be redefined to return the number of init stages required e g for a two stage init numInitStages should return 2 and initialize int stage must be implemented to handle the stage 0 and stage 1 cases The callInitialize function performs the full multi stage initialization for that module and all its submodules If you do not redefine the multi stage initialization functions the default behavior is single stage initialization the default numInitStages returns 1 and the default initialize int stage simply calls initialize end of simulation event The task of finish is solved in many simulators e g OPNET by introducing a special end of simulation event This is not a very good practice because the simu
325. t a long ago Changing it now would take quite some work and after all it works Member functions based on forEach cObject findObject const char s bool deep true Finds the object with the given name in a container object and returns a pointer to it or NULL if the object hasnt been found If deep is false only objects directly contained will be searched otherwise the function searches the whole subtree for the object bool ishe_there cObject obj Returns t rue if the given object obj is contained in this object s subtree 15 5 Module classes cModule cCompoundModule cSimpleModule Header file cmodule h sim directory There are three module classes cModule is the common base class for all module classes It contains data and functions to manage the module hierarchy in a network and handles parameters and gates cModule is used only as a base class no instance of cModule is ever created cCompoundModule is derived from cModule and represents compound modules in networks cCompoundmod has no added functionality over cModule cSimpleModule is also a descendant of cModule and is a base class for all simple module classes cObject cCoroutine cSimpleModule cCCompoundModule simple module classes 15 5 1 class cModule cModule contains data and functions to manage the module hierarchy in a network It also stores and controls module parameters and gates Construction destruction
326. t for a cell s be the total observation count in a cell n i plus the observation counts in all its sub sub sub etc cells and m the mother observations propagated to the cell We are interested in the n i m estimated amount of observations in the tree nodes especially in the leaves In other words if we have estimated observation amount Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 132 in a cell how to divide it to obtain m i1 m i2 M ik that can be propagated to child cells Naturally m i1 M i2 M ik i sa e E a h A ay seed E 8 TEM L Z o cee 2 Pul optimal when the s j values are very small and proportional distribution is good when the s __ jj values are large compared to m In practice a linear combination of them seems appropriate where A 0 means even and A 1 means proportional distribution Si i m OHA e Ah peer Ae 0 1 gt fioo 4 o 1 5 A r Figure 5 8 Density estimation from the k split cell tree We assume A 0 i e we distribute mother observations evenly Note that while n are integers m and thus are typically real numbers The histogram estimate calculated from k split is not exact because the frequency counts calculated in the above manner contain a degree of estimation themselves This introduces a certain cell division error the A parameter should be selected so th
327. t is not yet documented here See the User Manual 15 7 Module gate class cGate Header file cmodule h sim directory cGate object are created and managed by modules the user typically does not want to directly create or destroy cGate objects However they are important if a simple module algorithm needs to know about its surroundings Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 228 15 7 1 class cGate Construction destruction copying These functions are not documented here since they are considered internal to the simulation library Redefined virtual functions virtual const char className Returns pointer to the class name string cGate Gate identity int size If the gate is in a gate array returns the size of the vector otherwise it returns 1 int index If the gate is in a gate array returns the gate s position in it otherwise it returns 0 void setIndex int sn int vs Specifies that the gate is at index sn in a gate array of size vs This function should not be directly called by the user char type Returns the gate s type I for input and O for output cModule ownerModule Returns a pointer to the owner module of the gate int id Returns the position of the gate in the vector of all gates of the module void setOwnerModule cModule m int g Specifies that the gate is owned by module m and
328. t need to worry about constructors destructor className function etc The user can derive the new class from an existing simple module class not only cSimpleModule add new data members and add redefine member functions as needed Define_Module CLASSNAME Define _Module_Like CLASSNAME INTERFACENAME The use of this macro allows the creation of a module simple or compound when only its name is available as a string The macro expands to the definition a cModuleType object and some functions it needs The NEDC compiler generates Define _Module lines for all compound modules However it is the user s responsibility to put Define Module lines for all simple module types into one of the C sources 15 4 Common properties for all classes cObject Header file cob ject h sim directory Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 215 cOb ject is the base class for almost all classes in the OMNeT library cObject provides a name member a dynamically allocated string and a number of virtual functions These functions e either provide a default behaviour mostly good for all derived classes e or they are expected to be redefined in all derived classes Read this section carefully if you plan to create new classes 15 4 1 class cObject Construction destruction copying cOb ject cObject obj Copy cons
329. t they schedule events for when the gate finishes its current transmission and send then e alternatively you can implement channels with simple modules active channels The approach of some other simulators Note that some simulators e g OPNET assign packet queues to input gates ports and messages send are buffered at the destination module or the remote end of the link until received by the destination module With that approach events and messages are separate entities that is a send operation includes placing the message in the packet queue and scheduling an event which will signal the arrival of the packet In some implementations also output gates have packet queues where packets wait until the channel becomes free available for transmission OMNeT gates don t have associated queues The place where the sent but not yet received messages are buffered is the FES OMNeT s approach is potentially faster than the above mentioned solution because it doesn t have the enqueue dequeue overhead and also spares an event creation The drawback is as mentioned above that changes to channel parameters do not take effect immediately 5 6 Coding conventions Here s a bunch of advice on how to write OMNeT models Some of them are rules of thumb saying if you program like that you re likely to have less trouble other conventions are aimed at making the models produced by the OMNeT community more consistent
330. tOwner cObject newowner Sets the owner pointer of the object See documentation of cHead for more information Operations to be used by container classes derived from cObject void takeOwnership bool tk bool takeOwnership Sets returns the flag which determines whether the container object should automatically take ownership of the objects that are inserted into it void take cObject obj The function called by the container object when it takes ownership of the obj object that is inserted into it Implementation ob j gt setOwner this void drop cObject obj The function called by the container object when obj is removed from the container releases the ownership of the object and hands it over to its default owner Implementation ob j gt setOwner obj gt defaultOwner void free cObject obj The function is called when the container object has to delete the contained object obj It the object was dynamically allocated by new it is deleted otherwise e g if it is a global or a local variable it is just removed from the ownership hierarchy Implementation if obj gt storage D delete obj else obj gt setOwner NULL Load store operations The load store operations are used when running OMNeT on multiple processors They are used to exchange data between two instances of OMNeT running in different processes and possibly Technical University of Budapest Dept of Telecommunicatio
331. tPhase opening connection ev lt lt connection request from lt lt sre lt lt n Lele setPhase connection alive le setPhase closing connection ths Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 138 Writing out informative messages at strategic points of the code is an effective way debugging 6 16 2 Watches You may want some of your int long double char etc variables to be inspectable in Tkenv and to be output into the snapshot file In this case you can create cWat ch objects for them with the WATCH macro int i WATCH i char c WATCH c Tkenv also lets you change the value of the WATCHed variables The WATCH macro expands to a dynamically created cWatch object The object remembers the address and type of your variable The macro expands to something like new cWatch i i You can also make a WATCH for pointers of type char or cObject but this may cause a segmentation fault if the pointer does not point to a valid location when Tkenv or snapshot wants to use it You can also set watches for variables that are members of the module class or for structure fields WATCH lapbconfig timeout Placement of WATCHes Be careful not to execute a WATCH statement more than once as each call would create a new cWatch object If you use activity the best place for WATCHes is the top of the activity
332. te int g Returns reference to the gate identified with its index g Returns NULL if the gate doesnt exist cGate amp gate const char s int sn 0 Returns reference to the gate specified by name and index if multiple gate Returns NULL if the gate doesn t exist int findGate const char s int sn 0 Returns index of the gate specified by name and index if multiple gate Returns 1 if the gate doesn t exist Module parameters int params Returns total number of the module s parameters cPar amp par int p Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 220 Returns reference to the module parameter identified with its index p Returns NULL if the object doesn t exist cPar amp par const char name Returns reference to the module parameter specified with its name Returns NULL if the object doesn t exist int findPar const char s Returns index of the module parameter specified with its name Returns 1 if the object doesn t exist cPar amp ancestorPar const char name Searches for the parameter in the parent modules up to the system module It the parameter is not found an error message is generated Initialize and finish functions void initialize void initialize int stage int numInitStages void finish TBD Display Strings void setDisplayString int type const char dispstr bool imme
333. te submodule with the given name and optional index and returns its pointer If the submodule was not found returns NULL cModule moduleByRelativePath const char path Finds a submodule potentially several levels deeper given with its relative path from this module The path is a string of module full names separated by dots If the submodule was not found returns NULL 15 5 2 class cCompoundModule Redefined member functions bool isSimple Returns false virtual void scheduleStart simtime_t t Calls scheduleStart recursively for all its immediate submodules This is used with dynamically created modules virtual void deleteModule Calls deleteModule for all its submodules and then deletes itself 15 5 3 class cSimpleModule cSimpleModule is derived from cModule and is a base class for all simple module classes Most important cSimpleModule has a virtual member function called activity that is to be redefined in descendants this is the module function All basic functions associated with the simulation such as sending and receiving messages are implemented as cSimpleMod s member functions The module functions are run as coroutines during simulation Coroutines are brought to cSimpleModule from another base class called cCoroutine Construction destruction copying The user normally doesn t create new modules directly so these member functions are not documented here Redefined member fun
334. ted for the list for the cHead object the operator can be used to step from one element of the list to the next one cIterator is not a descendant of cObject cIterator cHead amp h Constructor cIterator will walk on the list passed as argument The current object will be the first one in the list void init cHead amp h Reinitializes the iterator object cObject operator Returns a pointer to the current object bool end Returns t rue if we reach the end of the list cObject amp operatort int Advances to the next object in the list 15 10 Non object container classes cLinkedList cBag There are two non object container classes cLinkedList and cBag The first one parallels with cQueue the second one with cArray 15 10 1 class cLinkedList cLinkedList is a container class that can holds non object items cLinkedList acts very similar to cQueue Memory management of contained items is controlled by the configPointer function As default pointers are treated as mere pointers so items are never duplicated or deleted Construction destruction copying cLinkedList cLinkedList amp llist Copy constructor Contained items that are owned by the list will be duplicated using the function passed in configPointer so that the new list will have its own copy of them By default there s no duplication function so only the pointers are copied cLinkedList const char name NULL
335. ter to a static string cMessage msg new cMessage NULL lt additional args gt const char str msg gt name gt returns ptr to fullName and fullPath Objects have two more member functions which return other sort of names based on the name attribute fullName and fullPath Suppose we have a module in the network university_lan compound module fddi_ring simple module station 10 If you call the functions on the simple module object cSimpleModule inherits from cOb ject too the functions will return these values ev lt lt module gt name gt station ev lt lt module gt fullName gt station 10 ev lt lt module gt fullPath gt university_lan fddi_ring station 10 Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 94 These functions work for any object For example a local object inside the module would produce results like this void FDDIStation activity cQueue buffer buffer ev lt lt buffer gt fullPath gt university_lan fddi_ring station 10 buffer fullName and fullPath together with className can be used for example to generate informative error messages Be aware that fullName and fullPath return pointers to static buffers Each call will overwrite the previous content of the buffer so for example you shouldnt put two ca
336. th simulators use a portable coroutine library proposed by Stig Kofoed based on setjmp longjmp On NT Parsec seemingly uses the thread implementation of the underlying operating system One advantage of OMNeT is that it can tell you how much stack space a module uses in fact see the stackUsage function so you can optimize the stack size Parsec apparently doesnt have this feature The worst problem with coroutines threads is that if you create too many of them youl need a hell lot of memory With the current engineering workstations it is practically impossible to create more Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix B Parsec and OMNeT 203 than a few times ten thousand entities in Parsec this is requires a few hundred megabytes of memory OMNeT has basically the same problem but it gives you two escapes 1 optimize the stack sizes by checking stackUsage 2 use handleMessage for some modules in very many cases it is more convenient anyway Parsec has no such escapes so the problem may hit you hard Comparison of PARSEC and OMNeT as parallel simulation tools A fundamental difference between Parsec and OMNeT is that Parser supports parallel simulation i e simulation on a multiprocessor while OMNeT supports distributed simulation over several hosts e If you do parallel simulation with Parsec the whole simulation will be a single operating system
337. the following components e Simulation kernel This contains the code that manages the simulation and the simulation class library It is written in C compiled and put together to form a library a file with a or lib extension Technical University of Budapest Dept of Telecommunications OMNeT Manual Overview 25 e User interfaces OMNeT user interfaces are used with simulation execution to facilitate debugging demonstration or batch execution of simulations There are several user interfaces written in C compiled and put together into libraries a or Lib files Simulation programs are built from the above components First the NED files are compiled into C source code using the NEDC compiler which is part of OMNeT Then all C sources are compiled and linked with the simulation kernel and a user interface to form a simulation executable Running the simulation and analyzing the results The simulation executable is a standalone program thus it can be run on other machines without OMNeT or the model files being present When the program is started it reads in a configuration file usually called omnetpp ini it contains settings that control how the simulation is run values for model parameters etc The configuration file can also prescribe several simulation runs in the simplest case they will be executed by the simulation program one after another The output of the simulation is written into data fil
338. the module hierarchy The surrounding compound module can be accessed by the parentModule member function cModule parent parentModule For example the parameters of the parent module are accessed like this double timeout parentModule gt par timeout cModule s findSubmodule and submodule member functions make it possible to look up the module s submodules by name or name index if the submodule is in a module vector The first one returns the numeric module ID of the submodule and the latter returns the module pointer If the submodule is not found they return 1 or NULL respectively CF int submodID compoundmod gt findSubmodule child 5 cModule submod compoundmod gt submodule child 5 The moduleByRelativePath member function can be used to find a submodule nested deeper than one level below For example compoundmod gt moduleByRelativePath child 5 grandchild would give the same results as compoundmod gt submodule child 5 gt submodule grandchild Provided that child 5 does exist because otherwise the second version will crash with an access violation because of the NULL pointer Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 110 The cSimulation moduleByPath function is similar to cModule s moduleByRelativePath function and it starts the search at the top level module Ite
339. the next message to segment B and announce it with the syncpoint call The simulation kernel sends the syncpoint to segment B When segment B s model time reaches the specified time segment B s simulation kernel blocks execution until the promised message arrives from A Then the Technical University of Budapest Dept of Telecommunications OMNeT Manual Parallel Execution 176 simulation continues typically but not necessarily with the message that has just been received from A In the reverse case when A is ahead of B A s message arrived at B before it has reached the syncpoint In this case there is no problem and the syncpoint is just an unnecessary precaution B just inserts the message in its future event set clears the syncpoint and continues execution The syncpoint API The syncpoint call takes two arguments The first is the model time when or more precisely when of after when the simple module will send a message to another simple module in a different segment The second argument is a gate given with its number or its name The gate implicitly specifies the destination segment to synchronize with syncpoint t outgate Details of the syncpoint implementation If the destination module is in the same segment the call is ignored This makes it possible to run models designed to execute in parallel as a single process without any modification Each segment keeps a list of syncpoints sent to it time
340. the object Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 231 TBD other set Value functions Retrieving the value const char stringValue j Returns value as char Only for string S type bool boolBalue Returns value as bool Converts from types Boolean B and long L long longValue Returns value as long Converts from types long L double D Boolean B function F distribution T and expression X double doubleValue Returns value as double Converts from types long L double D function F Boolean B distribution T and expression X cObject objectValue Returns value as pointer to cOb ject Type must be pointer O void pointerValue Returns value as pointer to cOb ject Type must be pointer P Other functions bool isNumeric Returns true if the stored value is of a numeric type cPar amp read Reads the object value from the ini file or from the user void convertToConst Replaces the object value with its evaluation a double Implemented as something like setValue D this gt doubleValue void configPointer VoidDelFunc delfunc VoidDupFunc dupfunc size_t size 0 Configures memory management for the void pointer P type Similar to cLinkedList s configPointer function delete func dupl func itemsize behaviour NULL NULL 0 Pointer is
341. there regardless of the state of the output vector object 6 14 2 Output scalars While output vectors are to record time series data and thus they typically record a large volume of data during a simulation run output scalars are supposed to record a single value per simulation run You can use outputs scalars e torecord summary data at the end of the simulation run e todo several runs with different parameter settings random seed and determine the dependence of some measures on the parameter settings For example multiple runs and output scalars are the way to produce Throughput vs Offered Load plots Output scalars are recorded with the recordScalar member function It is overloaded you can use it to write doubles and strings const char double avg_throughput total_bits simTime recordScalar Average throughput avg_throughput You can record whole statistics objects by calling recordStats cStdDev eedstats new cStdDev recordStats End to end Statistics eedstats Calls to recordScalar and recordStats are usually placed in the redefined finish member function of a simple module The above calls write into the textual output scalar file The output scalar file is preserved across simulation runs unlike the output vector file is scalar files are not deleted at the beginning of each run Data are always appended at the end of the file and output from different simulation runs ar
342. tion Library 96 The opp_concat function might also be useful for example in constructing object names It takes up to four const char pointers concatenates them in a static buffer and returns pointer to the result The result s length shouldn t exceed 255 characters 6 3 Messages and packets 6 3 1 The cMessage class In OMNeT 4 cMessage is a central class Objects of cMessage and subclasses may model a number of things events messages packets frames cells bits or signals travelling in a network entities travelling in a system and so on Attributes A cMessage object has number of attributes Some are used by the simulation kernel others are provided just for the convenience of the simulation programmer A more or less complete list e The name attribute is inherited from cObject e The message kind attribute is supposed to carry some message type information Zero and positive values can be freely used for any purpose Negative values are reserved for use by the OMNeT simulation library especially MK_PACKET 1 and MK_INFO 2 are used to denote that the message is a network packet see the cPacket class later e The length attribute understood in bits is used to compute transmission delay when the message travels through a connection that has an assigned data rate e The bit error flag attribute is set to true by the simulation kernel with a probability of 1 1 ber when the message is sent through a c
343. tion about module activity eee eeeseeeeeteceseeeeeeeeneees 137 G16 2 W aches Rc RG CA BEN Oe GSR BER 138 616 3 HAP SUIS recan a a a N a os ecaad E E a 138 GV G H TEA 0I AEA E T A E a Sheanaroaees 142 6 16 5 Disabling Waring S iacsicnn aaa E E E a 142 6 16 6 Getting coroutine stack usage sseesssesseesseeesseeessetsseessersseresseeessresseesseessees 142 6 17 Changing the network graphics at run time seesssssessessseeeseetesstesseessessserssseeessresseesse 143 6 18 Tips for speeding up the simulation s ssssssesssesesssetssressesssesessetssstesseessersseeesseessseesseesse 143 Technical University of Budapest Dept of Telecommunications OMNeT Manual Introduction 8 O18 Wsine shared ODjEC tS jeccsusassouncess aes n aes ate teatecue ia a 144 6 19 Building large networks adeno een nre ae maaan eres 145 6 419 1 Generating NED Tiles sissies etd nell e oleae aie ete 145 6 19 2 Building the network from C code eee eeceeseeseceeeeeeeeesaeecnaecnseenseeeeaeees 146 7 BUILDING SIMULATION PROGRAMS sssissscosssseassosessciesesdessensecssuseasesasoosesscssnastsstcstade 147 Dek OVCEVIEW oseni i E E A e E A enlist a E 147 12 Using Unix and 26s scccsscadaa cs assiata siren a R E Ra R E E ES 148 A Del Installation inneni eia ia iiai akes 148 7 2 2 Producing a makefile with the makemake script seeeeeeeeeeeeeeeseeeseseeesrreresresse 149 1 2 3 Multi directory models isvccsccxsacsieisoesadcassaccave iron
344. tions topology description Simulation methodology Precompiler C support alternative to coroutines threads complex message receiving constructs message types module gates sending via gates module parameters dynamic module entity creation Simulation library statistics histogram classes routing support FSM support support for output files container classes Parallel execution conservative optimistic statistical synchronization yes optionally delay data rate bit error rate via NED nedc no need code is standard C based on C handleMessage timeout only via subclassing cMessage or via cMessage pars yes yes yes also compound modules yes cStdDev 3 histogram classes P k split yes cTopology yes FSM macros yes cOutVector recordScalar yes cQueue cArray yes explicit support 204 manually from the driver entity pcc Parsec compiler language based on C yes filter by message type timeout guards etc via the message construct direct sending only yes yes much more elaborate than in OMNeT possible but no support Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix B Parsec and OMNeT 205 13 5 Correspondence between Parsec and OMNeT scheduleAt message simTime delay n a send message gate Parsec has no equ
345. tions which are called from activity and moving the some local variables of activity into the module class so that they can also be accessed by the new member functions The above approach doesn t work in Parsec because Parsec is C based and entities are not C classes Of course one may call ordinary C functions from the entity body but the necessary parameters must be passed in the argument list or as pointers to data structures in the entity Another solution in Parsec is to use a construct called friend functions not to be confused with C friend functions Parsec s friend functions may access the local variables of the entity quite strange in C but much like an inner procedure in Pascal However the Parsec documentation does not recommend using friend functions they are slow it says they are provided for Maisie compatibility Message sending Messages can be sent to other entities with the send construct send message to dest entity after delay For example creating a new message of type Request with the parameters 10 and self the current entity and sending it to ent it y2 entity after a delay looks like this send Request 10 self to entity2 after 5 Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix B Parsec and OMNeT 201 This Parsec construct is totally equivalent in functionality to OMNeT s sendDirect message delay dest module dest gate call Since P
346. to be written one must explicitly enable parameter change logging with the log parchanges option The load libs entry can be used to load shared objects containing simple modules compiled NED code etc at run time Each setting has a meaningful default value Almost any of the above options can also be specified individually for each run Per run settings if they exist have priority over globally set one 8 2 2 Splitting up the configuration file OMNeT supports file inclusion in ini files This feature allows you to partition large ini files to logical units fixed and varying part etc An example omnetpp ini include parameters ini include per run pars ini Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 158 8 2 3 Module parameters in the configuration file Values for module parameters go into the Parameters or the Run 1 Run 2 etc sections of the ini file The run specific settings take precedence over the overall settings Parameters that were assigned a non input value in the NED file are not influenced by ini file settings Wildcards can be used to supply values to several model parameters at a time Filename style glob and not regex style pattern matching is used Character ranges use curly braces instead of square brackets to avoid interference with the notation of module vectors a zA Z If a parameter name matches several wildcard
347. to the message An example cLongHistogram pklen_distr new cLongHistogram pklen_distr msg gt parList add pklen_distr cLongHistogram pklen_distr cLongHistogram msg gt parList get pklen_distr You should take care that names of the attached objects do not clash with parameter names If you do not add parameters to the message and do not call the parList function the internal cArray object will not be created This saves you both storage and execution time You can attach non object types or non cOb ject objects to the message by using cPar s void pointer P type see later in the description of cPar An example struct conn_t conn new conn_t conn_t is a C struct msg gt addPar conn void conn msg gt par conn configPointer NULL NULL sizeof struct conn_t Runtime overhead It has been reported that using cPar message parameters might account for quite a large part of execution time sometimes as much as 80 If your simulation is going to be very CPU intensive you re probably better off subclassing either cMessage or rather cPacket and adding the required parameters as ints Longs bool1s etc to the new message class Some time in the future OMNeT will directly support message subclassing and it will make the new parameters inspectable in the graphical user interface Tkenv This is a feature much demanded by users However if you don t expect yo
348. togram amp res Duplication and assignment work all right with cDoubleHistogram Redefined virtual functions virtual const char className Returns a pointer to the class name string CDoubleHistogram virt virt ual void info char buf ual const char inspectorFactoryName Redefined virtual void writeContents ostream amp os Writes the object contents to the stream virtual void transform If the result collection is in the first phase transforms the values into histogram If already in the second phase the function does nothing virtual unsigned cells Returns the number of cells virtual double basepoint unsigned k Returns the kth basepoint virtual unsigned cell unsigned k Returns the number of samples that fell into the kth subinterval virtual double cellPDF unsigned k Returns the calculated PDF in the kth subinterval virtual double random Returns a random number based on the distribution collected If no values have been collected it returns 0 when in initial collection phase it returns one of the stored observations after the histogram has been set up a random integer is returned virtual void clearResult Clears the distribution collected so far 15 13 The P and k split algorithms cPSquare cKSplit Header file cposquare h cksplit h sim directory 15 13 1 class cPSquare Implements the P algorithm Technical Univer
349. trates dynamic module creation The FDDI example is an accurate FODDI MAC simulation which was written on the basis of the ANSI standard The following table summarizes the sample simulations NAME TOPIC DEMONSTRATES nim a simple two player game module inheritance module type as parameter hcube hypercube network with hypercube topology with dimension as parameter deflection routing topology templates output vectors token Token Ring network ring topology with the number of nodes as parameter using cQueue wait and the putaside queue output vectors fifol single server queue simple module inheritance decomposing activity into several functions using simple statistics and output vectors printing stack usage info to help optimize memory consumption Technical University of Budapest Dept of Telecommunications OMNeT Manual An Example The NIM Game using finish 34 ifo2 another fifo implementation FDDI MAC simulation using handleMessage decomposing handleMessage into several functions the FSM macros simple module inheritance using simple statistics and output vectors using finish using statistics classes and many other features hist demo of the histogram classes collecting observations into statistics objects saving Statistics objects to file and restoring them using the inspect 1st file in Tkenv dyna a client server network
350. tructor In derived classes it is usually implemented as operator ob cObject const char s cObject h Creates a cObject with the given name The owner will be the h object if the pointer is not NULL that is the constructor contains asetOwner h call virtual cObject Virtual destructor Deletes the name and notifies the user interface that the object has been destructed virtual cObject dup Duplicates the object and returns a pointer to the new one In derived classes it is usually implemented as return new cObject this cObject amp operator cObject amp o The assignment operator Most derived classes contain a cSomeClass amp cSomeClass operator cSomeClass amp function The assignment operators do not copy the object name If you want to do so you can copy it be hand setName o name void operator new size_t m cObject s operator new does more than the global new It cooperates with cOb ject s constructor to determine the storage class of the object static auto or dynamic char storage Returns the storage class of the object The return value is one of the characters S A D which stand for static auto and dynamic respectively void destruct Direct call to the virtual destructor Name member virtual void setName const char s Sets object s name The object creates its own copy of the string NULL pointer may also be passed const char name Return
351. ts The time when events occur is often called event timestamp with OMNeT well say arrival time because in the class library the word timestamp is reserved for a user settable attribute in the event class Time within the model is often called simulation time model time or virtual time as opposed to real time or CPU time or which refers to how long the simulation program has been running or how much CPU time it has consumed 5 1 2 The event loop Discrete event simulations maintain a set of future events in a data structure often called FES Future Event Set Such simulators usually work according to the following pseudocode Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 62 initialize this includes building the model and inserting initial events to FES while FES not empty and simulation not yet complete retrieve first event from FES t timestamp of this event process event processing may insert new events in FES or delete existing ones finish simulation write statistical results etc The first initialization step usually builds the data structures representing the simulation model calls any user defined initialization code and inserts initial events into the FES to ensure that the simulation can start Initialization strategy can be quite different from one simulator to another The subsequent loop consumes events from the FES and processes them Events ar
352. ty to cSimpleModule This section discusses cSimpleModule s four previously mentioned member functions intended to be redefined by the user initialize activity handleMessage and finish 5 3 1 activity Process style description With activity you can code the simple module much like you would code an operating system process or a thread You can wait for an incoming message event at any point of the code you can suspend the execution for some time model time etc When the activity function exits the module is terminated The simulation can continue if there are other modules which can run The most important functions you can use in activity are they will be discussed in detail later e receive family of functions to receive messages events e wait to suspend execution for some time model time e send family of functions to send messages to other modules e scheduleAt to schedule an event the module sends a message to itself e cancelEvent to delete an event scheduled with scheduleAt e end to finish execution of this module same as exiting the activity function The activity function normally contains an infinite loop with at least a wait or receive call in its body Examples TBD Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 70 Application area One area where the process style descri
353. type can occur in the same network description or in an imported network description Module vector as submodule It is possible to create an array of submodules a module vector This is done with an expression between brackets right behind the module type name The expression can refer to module parameters A zero value as module count is also allowed Example module BigCompound parameters num_of_submods const submodules Submodl1 Node 3 PP ees Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 40 Submod2 Node num_of_submods EL wes Submod3 Node num_of_submods 1 2 Ef aers endmodule Module type as parameter Instead of supplying a concrete module type one can leave it as a parameter At the same time to let the NED compiler know what parameters and gates that module has the user has to supply the name of an existing module type This is done with the 1ike phrase Example module CompoundModule parameters node_type string gates submodules theNode node_type like GeneralNode parameters buffer 10 connections endmodule The above example means that the type of the submodule t heNode is not known in advance it will be taken from the node_type parameter of CompoundModule which must be a string for example SwitchingNode The module type called GeneralNode must have appeared earlier in the NED files its declaration will be used t
354. ue queue initialize data member Stack size decides between activity and handleMessage e if the specified stack size is zero haandleMessage will be used e if itis greater than zero activity will be used Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 68 If you make mistake e g you forget to set zero stack size for a handleMessage simple module the default versions of the functions issue error messages telling you what is the problem 5 2 5 Decomposing activity handleMessage and inheritance It is usually a good idea to decompose a activity or handleMessage function when it grows too large Too large is a matter of taste of course but you should definitely consider splitting up the function if it is more that a few screens say 50 100 lines long This will have a couple of advantages e will help future readers of the code understand your program e will help you understand what it is you re really programming and bring some structure into it e will enable you to customize the class by inheriting from it and overwriting member functions If you have variables which you want to access from all member functions typically state variables are like that you ll need to add those variables to the class as data members Let s see an example class TransportProtocol public cSimpleModule public Module_Class_Members TransportProtocol c
355. ulation if the function is called frequently bool isThereMessage Tells if the next message in the event queue is for the same module and has the same arrival time Returns t rue only if two or more messages arrived to the module at the same time Getting messages from the put aside queue or the event queue cMessage receive simtime_t t MAXTIME Returns the first message from the put aside queue or if it is empty calls receiveNew to return a message from the event queue with the given timeout Note that the arrival time of the message returned by receive can be earlier than the current simulation time cMessage receiveOn const char s int sn 0 simtime_t t MAXTIME Scans the put aside queue for the first message that has arrived on the gate specified with its name and index If there is no such message in the put aside queue calls receiveNew to return a message from the event queue with the given timeout Note that the arrival time of the message returned by receive can be earlier than the current simulation time cMessage receiveOn int g simtime_t t MAXTIME Same as the previous function except that the gate must be specified with its index in the gate array Using this function instead the previous one may speed up the simulation if the function is called frequently Waiting void wait simtime_t time Wait for given interval The messages received meanwhile are insert
356. ur at the same simulation time Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 103 handle msg as an old message The correct way to do this is to check the putaside queue bool queue_was_empty putAsideQueue empty cMessage msg receive if queue_was_empty handle msg as an old message To discard the contents of the put aside queue one could use the following code while putAsideQueue empty delete receive To demonstrate receiveoOn the following code fragment waits for a message on one specific input gate and discards all messages that arrived on other gates in the meanwhile cMessage msg receiveNewOn important_input_gate while putAsideQueue empty delete receive The above code is almost equivalent to the following except that it preserves the previous contents of the put aside queue cMessage msg for msg receiveNew if msg gt arrivedOn important_input_gate break delete msg All message receiving functions can be given a timeout value This is a delta not an absolute simulation time If an appropriate message doesn t arrive within the timeout period the function returns a NULL pointer An example simtime_t timeout 3 0 cMessage msg receive timeout if msg NULL timeout expired without any messages else process message 6 4 5 The wait funct
357. ur simulations execute for hours then cPar parameters are the most convenient way to go Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulation Library 99 6 3 3 Message encapsulation It is often necessary to encapsulate a message into another when youre modeling layered protocols of computer networks Although you can encapsulate messages by adding them to the parameter list there s a better way The encapsulate function encapsulates a message into another one The length of the message will grow by the length of the encapsulated message An exception when the encapsulating outer message has zero length OMNeT assumes it is not a real packet but some out of band signal so its length is left at zero cMessage userdata new cMessage userdata userdata gt setLength 8 2000 cMessage tcpseg new cMessage tcp tcpseg gt setLength 8 24 tcpseg gt encapsulate userdata ev lt lt tcpseg gt length lt lt endl gt 8 2024 16192 A message can only hold one encapsulated message at a time The second encapsulate call will result in an error It is also an error if the message to be encapsulated isn t owned by the module You can get back the encapsulated message by decapsulate cMessage userdata tcpseg gt decapsulate decapsulate will decrease the length of the message accordingly except if it was zero If the length w
358. usses each of these types in detail 4 1 2 Reserved words The writer of the network description has to take care that no reserved words are used for names The reserved words of the NED language are import include channel endchannel simple endsimple module endmodule error delay datarate const parameters gates submodules connections gatesizes on if machines for do endfor network endnetwork nocheck ref ancestor true false like input numeric string bool char 4 1 3 Case sensitivity The network description and all identifiers in it are case sensitive 4 2 The import statement Example import tkn_mod tkn2_mod Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 36 The import statement the include keyword is also recognized for backwards compatibility is used to import declarations from other network description files After importing a network description one can use the components channels simple compound module types defined in it From the imported files only the declaration information is used but no C code is generated The consequence is that one has to compile and link each network description not only the top level ones The user can specify the name of the files with or without the ned extension One can also include a path in the filenames or better use the NEDC compilers I lt path gt command line option to name the directories where the imported f
359. ut gate of the compound module and the destination gate can be an input gate of a submodule or an output gate of the compound module If the user does not specify a channel the connection will have no propagation delay no transmission delay and no bit errors Sender outgat gt Receiver ingate The arrow can point either left to right or right to left The user can also specify a channel by its name Sender outgate gt Dialup1l4400 gt Receiver ingate In this case the NED sources must contain the definition of the channel One can also specify the channel parameters directly Sender outgate gt error le 5 delay 0 001 gt Receiver ingate Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 44 Either of the parameters can be omitted and they can be in any order Loop connections If submodule or gate vectors are used it is possible to create more than one connection with one statement This is termed a multiple or loop connection A multiple connection is created with the for statement for i 0 4 do Sender outgate i gt Receiver i ingate endfor The result of the above loop connection can be illustrated as follows E E E E E re Coe E Com Fig 4 1 Loop connection One can place several connections in the body of the for statement separated with semicolons More than one indices can be specified in a for statement with their own lower and u
360. utable This architecture enables the GUI to be very powerful every user created object is visible and modifiable in the GUI via inspector windows and the user has tight control over the execution To the author s best knowledge the tracing feature OMNeT provides is unique among the C based simulation tools Performance What performance can be expected from the simulation Simulation programs typically run for several hours Probably the most important factor is the programming language almost all network simulation tools are C C based Performance is a particularly interesting issue with OMNeT since the GUI debugging tracing support involves some extra overhead in the simulation library However in a reported case an OMNeT simulation was only 1 3 slower than its counterpart implemented in plain C i e one containing very little administration overhead which is a very good showing A similar result was reported in a performance comparison with a Parsec simulation Source Availability J s the simulation library available in source This is a trivial question but it immediately becomes important if one wants to examine or teach the internal workings of a simulation kernel or one runs into trouble because some function in the simulation library has a bug and or it is not documented well enough In general it can be said that non commercial tools like OMNeT are open source and commercial ones are not This is also true for OPNET t
361. ve to set the variable yourself 7 4 4 Creating project files for your simulations 1 Start by copying amp renaming one of the dsp files from the samples directory It already contains the Tkenv Cmdenv configurations etc 2 Rename all cc files to cpp ren cc cpp and add them to the project 3 Add the ned files to the project and set custom build option for them Description NED Compiling InputPath Command nedc s _n cpp InputPath Outputs S InputPath _n cpp Hint you can select all ned files together and All configurations from the combo at the left of the Settings dialog and then you have to type this settings only once 4 For each ned file add a corresponding _n cpp file Hint if you compile the ned files choose Compile from the menu the _n cpp files will be created and you can Select them all at once in the Add files dialog 5 Make sure to turn off exception handling and RTTI they interfere with the coroutine library and set the necessary reserved stack size 6 Note for Tkenv link with sim_std lib envir lib tkenv 11ib and the Tcl Tk libraries link as Win32 Console app For Cmdenv you need to link with sim_std lib envir 1lib and cmdenv lib It is planned to create wizards in the future to ease some of these steps Technical University of Budapest Dept of Telecommunications OMNeT Manual Building Simulation Programs 154 7 4 5 Reconfigu
362. very slow to program in Motif 8 6 Typical problems 8 6 1 Stack problems Stack violation FooModule stack too small in module bar foo OMNeT detected that the module has used more stack space than it has allocated You should increase the stack for FooModule You can call the stackUsage from finish to find out actually how much stack the module used Error Cannot allocate nn bytes stack for module foo bar If you get the above message you have to increase the total stack size the sum of all coroutine stacks You can do so in omnetpp ini General total stack kb 2048 2MB There is no penalty if you set total stack kb too high I recommend to set it to a few K less than the maximum process stack size allowed by the operating system ulimit s see next section Segmentation fault On Unix if you set the total stack size higher you may get a segmentation fault during network setup or during execution if you use dynamically created modules for exceeding the operating system limit for maximum stack size For example in Linux 2 0 x the stack can be at most 8192K that is Technical University of Budapest Dept of Telecommunications OMNeT Manual Running The Simulation 169 8MB The ulimit syscall and utility program can be used to modify the resource limits but you can only increase if you re root Furthermore resource limits are inherited by child processes The following statement worked
363. vir printf but just writes out its argument string with no formatting bool askf char buf int len const char profmt Pops up a dialog displays the message given in profmt and following arguments in printf format and reads a line maximum len characters from the user into the buffer buf Returnst rue if the user pressed the Cancel button bool gets const char prompt char buf int len 255 Similar to cEnvir askf but just writes out the prompt message string with no formatting bool askyesno const char msgfmt Puts a yes no question to the user The question itself is expected in the printf format format string arguments The t rue return value means yes false means no cEnvir amp prompt const char s j Sets the prompt for subsequent gt gt operators There is only one global prompt text that is shared between all simple modules so between setting the prompt string and asking the user no receive peekmsg messages or wait call is recommended Or the user may experience funny things const char prompt Returns the prompt string Technical University of Budapest Dept of Telecommunications OMNeT Manual Appendix D Class Library Reference 255 Overloaded operators Overloaded operators provide iostream like I O for cEnvir inline cEnvir amp operator lt lt cEnvir ev const s
364. virtual void activity this redefined virtual function holds the algorithm Define _Module Game operation of Game void Game activity strings to store player names player 0 is unused char player 3 32 read parameter values int num_sticks par num_sticks int player_to_move par first_move waiting for players to tell their names for int i 0 i lt 2 i cMessage msg receive if msg gt arrivedOn from_playerl strcpy player 1l msg gt name else strcpy player 2 msg gt name delete msg ev represents the user interface of the simulator ev lt lt Let the game begin n ev lt lt Player 1 lt lt player 1 lt lt Player 2 lt lt player 2 lt lt Anin do ev lt lt Sticks left lt lt num_sticks lt lt n ev lt lt Player lt lt player_to_move lt lt lt lt player player_to_move lt lt to move n cMessage msg new cMessage num_sticks num_sticks will be the msg kind if player_to_move 1 send msg to_playerl else send msg to_player2 msg receive int sticks_taken msg gt kind delete msg num_sticks sticks_taken ev lt lt Player lt lt player_to_move lt lt lt lt player player_to_move lt lt took lt lt sticks_taken lt lt stick s n Technical University of Budape
365. with the usual math operators they can use parameters taken by value or by reference call C functions contain random and input values etc 4 9 1 Using parameters in expressions ref and ancestor Expressions can use the parameters of the compound module being built A parameter can be taken by value or by reference The default is by value to select by reference passing for a parameter you have to use the ref modifier Parameters passed by reference can be used by a module to propagate values status info etc to other modules The ancestor modifier allows one to access parameters from higher in the module hierarchy module Compound parameter nnn submodules proc Processor parameters parl ref nnn 2 par2 10 ancestor par_somewhere_up endmodule 4 9 2 Operators The following operators can be used in expressions in order of precedence operator meaning asi la Y unary minus negation bitwise complement power of Kf g S multiply divide modulus tS add subtract lt lt gt gt bitwise shifting amp bitwise and or xor is reserved for power equal t not equal gt gt greater greater or equal lt lt less less or equal amp amp lIl logical operators and or xor 2i the C C inline if Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 54 4 9 3 The si
366. written out If you run the simulation with Tkenv you can also create a snapshot from the menu An example of a snapshot file SNAPSHOT Of object simulation Label three station token ring Sim time 0 0576872457 57ms Network token Run no L Started at Mar 13 1997 14 23 38 Time Mar 13 1997 14 27 10 Elapsed D5 SEC Initiated by operator cSimulation simulation begin Modules in the network token 1 TokenRing comp 0 2 Comp mac 3 TokenRingMAC gen 4 Gene sink 5 Sink p e e uter 4 comp 1 6 Computer mac 7 TokenRingMAC gen 8 Generator sink 9 Sink comp 2 10 Computer mac 11 TokenRingMAC gen 12 Generator sink 13 Sink end cCompoundModule token begin 1 params cArray n 6 1 gates cArray empty comp 0 cCompoundModule 2 comp 1 cCompoundModule 6 comp 2 cCompoundModule 10 end cArray token parameters begin num_stations cModulePar 3 L num_messages cModulePar 10000 L ia_time cModulePar truncnormal 0 005 0 003 F THT cModulePar 0 01 D data_rate cModulePar 4000000 L cable_delay cModulePar 1e 06 D end cModulePar token num_stations begin Type L Value 3 end Technical University of Budapest
367. y name each time like d par delay you may query its pointer into a cPar state variable with the same name prepended with p like pDelay amp par delay Use ev printf and ev lt lt see later to print out information on what the module is doing Doing so will pay out several times when it comes to debugging Use a parameter and a state variable called debug Surround your debugging output ev lt lt and ev printf calls with if debug You may introduce more specific debug switches like debug_queuing etc Technical University of Budapest Dept of Telecommunications OMNeT Manual Simple Modules 87 5 7 Component libraries Because of the structure of the simulation system one can create libraries of reusable elements in several ways The three basic types are e simple module libraries e NED source libraries e precompiled compound module libraries The elegant thing is that the user of the library does not need to know which kind of library he she is using the three types of libraries are equivalent in terms of usage 5 7 1 Simple module libraries Simple modules that can be used in more than simulations can form an object library Good candidates for module libraries are simple modules that implement e Physical Data link protocols Ethernet Token Ring FDDI LAPB etc e Higher layer protocols IP TCP X 25 L2 L3 etc e Network application types E mail NFS X audio etc e Basic element
368. ze of outvec index outvecSgate gt index gt 5 it is gate 5 in the vector int int For non vector gates size returns 1 and index returns 0 The type member function returns a character I for input gates and O for output gates char type outgate gt type gt 0 Gate IDs Module gates input and output single and vector are stored in an array within their modules The gate s position in the array is called the gate ID The gate ID is returned by the id member function int id outgate gt id For a module with input gates from_app and in 3 and output gates of to_app and status the array may look like this dir name index input from_app output to_app empty input in 0 input LAEL input in 2 output status The array may have empty slots Gate vectors are guaranteed to occupy contiguous IDs that is it is legal to calculate the ID of gate k as gate gate 0 id k Message sending and receiving functions accept both gate names and gate IDs the functions using gates IDs are a bit faster Gate IDs do not change during execution so it is often worth retrieving them in advance and using them instead of gate names Gate IDs can also be determined with the findGate member of cModule int idl findGate out int id2 findGate outvect 5 Technical University of Budapest Dept of Telecommunications OMNeT Manual The Simulatio
369. zeof and index operators A useful operator is sizeof which gives the size of a vector gate The index operator gives the index of the current submodule in its module vector An example for both module Compound gates in fromgens submodules proc Processor sizeof fromgens parameters address 10 1 index connections for i 0 sizeof fromgens 1 do in i gt proc i input endfor endmodule Here we create as many processors as there are input gates for this compound module in the network The address parameters of the processors are 10 20 30 etc 4 9 4 Time constants Anywhere you would put numeric constants integer or real to mean time in seconds you can also specify the time in units like milliseconds minutes or hours parameters propagation_delay 560ms 0 560s connection_timeout 6m 30s 500ms 390 5s lunchtime 0 5h 30 min The following units can be used 10 10 10 60 3600 60 3600 4 9 5 Random values OMNeT has the following predefined distributions e uniform uniform integer e exponential e normal truncated normal Technical University of Budapest Dept of Telecommunications OMNeT Manual The NED Language 55 Each distribution has one or more parameters Examples uniform 0 1 uniform in 0 1 intuniform 2 2 uniform int limits included 2 1 0 1 or 2 exponential 5 exponential with mean 5

OMNeT++ - Fossies

Contents

Download Pdf Manuals

Related Search

Related Contents