Home

PlaSMA User Guide - Platform for Simulations with Multiple Agents

Contents

1. Figure 5 2 Excerpt from the build configuration template file configuration template S Pasma PlaSMA Scenario Template configuration properties Eclipse Platform Se File Edit Source Refactor Navigate Search Project Run Window Help ri 0 Q gt EEO OOF PIP HrEreers es E Plasma 5 Navigator Package Explorer 23 E T scenario_implementat El configurationtempla El configuration proper 3 configuration proper 2 Of Ant E3 gt 52 Outline El Task List 0 ESB 7 SCENARIO USER CONFIGURATION FOO x amp Eg PlaSMA Core trunk i an AN marica PlaSMA Core Application Launcher Gi PlaSMA Scenario B4 Common scenarios scenario_ e A en PlaSMA Core Build Management lag PlaSMA Scenario Germany scenarios scenario_germ AA Choose the buildfiles to add PlaSMA Scenario B4 Common Codebase ey PlaSMA Scenario Learning Warden scenarios scenari 2 gt El PIaSMA Core E PlaSMA Scenario Germany E PlaSMA Scenario Northern Germany scenarios sceni For windows user El PlaSMA Scenario B4 Common le PlaSMA Scenario Northern Germany eq PlaSMA Scenario Template scenarios scenariotemp pasa instailat p 22 PlaSMA Scenario Germany dd PlaSMA Scenario Test PlaSMA Worldmodel GG sre with double bac b EY PlaSMA Scenario Learning Warden M JRE System Library jdk1 6 0 13 plasma location 22 PlaSMA Scenario Northern Germany Ea LS
2. S9 PlaSMA eS Pr Multiagent Simulation PlaSMA Multiagent Simulation User Guide Version 13 03 March 2013 Contents Contents iii 1 Introduction 1 2 PlaSMA Setup 3 2 1 Installation and Setup of required and optional Software 3 2 1 1 PlaSMA Requirements o 3 2 2 PIlaSMA Server Configuration 2 a e 10 2 3 PlaSMA Client Configuration c s o o sopot booe ea aeee E E P e 12 3 Running PlaSMA 15 3 1 PlaSMA Simulation Server lt lt e e raa sss seserii reto ra FE VEAN 15 SII Server States cei a aa a A a A AAA 15 3 1 2 Parallel and Distributed Simulation lt o e ecne mee ee 16 321 PESMA Cl a araia dese cars I E eee a E a 0 aet oe agea AE ee is 17 3 2 1 VisualClient ee aa a ao o oe ee T e A 17 3 2 2 Command Line Interface sed cadorna emanic aa a a ar e e a 18 3 3 Error Handling s anii sei diep ap ech a a a a a i eR BR Re Bo EO ai t 20 4 Scenario Configuration 21 4 1 General Simulation Settings cced pew a e E aa a 21 4 1 1 Synchronisation Configuration 2 2 0 ee 22 4 2 Domain Gontiguration 242 4 cn4 a eee aed PARE DEERE BO 23 4 2 1 Scenario Ontology Specification 2 ooa ee 23 4 2 2 Simulation Agents oa oaa 23 423 World Model ssa ie d 5 4842 8 6 08 2 e oi ew doh A 24 4 2 4 Online Scenario Visualization 2 2 ee 25 43 Simulation Logging q pts add a Ree ee ek BRO EE 26 43 1 Log Levels sac a eae eee eb eee ee we ee 26 4 3 2 hopping Categories 2 d a o a
3. 5 4 World Model Design A simulation usually consists of an entity or multiple entities that represents the physical environment in the simulation In the case of PlaSMA this environment is provided by a dedicated world model which stores position and state of every physical object Additionally the world model provides a graph based infrastructure where vertices represent locations and edges represent connections between them such as roads or railways The PlaSMA world model is created from an ontology specification using W3C Web Ontology Language OWL BvHH 04 PSHHO04 in its OWL DL sub language Ontologies are specified in ow files with XML encoding While there are several predefined ontologies for object types in logistic scenarios see Section 5 4 a new scenario needs its own instance declarations stated in a scenario specific ontology file see Section 5 4 3 The following subsections provide the technical background of ontologies their usage for PlaSMA world models as well as how to create own scenario ontologies Technical Background Ontology knowledge bases are mainly separated in two distinct parts TBox and ABox While the TBox defines the schema level the ABox specifies instances But this distinction is not necessarily reflected in separate ontology files The TBox defines terminological knowledge i e it incorporates concept definitions and property definitions Concepts or classes describe the schema of everyt
4. Windows server master bat lt password gt jade lt jade args gt The specification of the database password is optional You can also specify the password in the XML server configuration lt plasma_dir gt etc serverconfig xml If the password is given as command line argument it also overrides the XML configuration It is required that the PlaSMA server and the database run on the same machine PlaSMA uses JADE as agent environment You can add JADE arguments by using the jade option All following arguments are passed to JADE 3 1 1 Server States A running PlaSMA server is in either of one of the following states e Blank The state right after start No scenario was loaded yet e Loading scenario The server is loading a simulation configuartion file e Scenario loaded The simulation has loaded a scenario configuration file 15 16 CHAPTER 3 RUNNING PLASMA e Initializing The server is initializing a simulation scenario by creating the initial world model and setting up all participating agents e Scenario ready Scenario initialization is done The simulation is ready to be started e Running The server is currently simulating e Paused The current simulation run is paused and can be resumed e Run Stopped The last simulation run has finished You can either trigger a new run of the same scenario or load another scenario e Error A critical error occurred on the server and the simulation cannot proceed T
5. a PlaSMA Scenario Template Gh config Furthermore ple gt build common D configxm plasma location C 6 confi E g Ea ow F buildxmi IB scenario owi TE che scenario l PlaSMA Scenario Test PlaSMA Worldmodel sa E build xml 1549 14 04 09 19 03 warden be deployed prop configuration properties configure the bu arios E configuration template 1276 18 05 08 15 18 warder YOU depend on al ing the TE PlaSMA Scenario Test PlaSMA Worldmodel scenario following line i Eu PlaSMA Userguide trunk doc userguide P ACERO apa STATIC CONFIGURA STATIC CONFI Only show xml files PlaSMA Core Libr plasma libs plas Scenario Build Si 5 cas O Cancel El Problems 2 Declaration El Console 23 4 Tasks Progress Ex 68 4 El 147 WO Javadoc 23 Texlipse running C Program FilesiMiKTeX 2 7 miktex bin pdflatex exe interaction nonstopmode src spe al m rife m r n Figure 5 3 Using the Ant build system for scenarios in the Eclipse IDE 32 CHAPTER 5 SCENARIO IMPLEMENTATION E Windows PowerShell V2 CTP ioj x PS C Users warden Subversion P1laSMA_Scenarios scenarios scenar1o_b4_common gt ant clean deploy Buildfile build xml cl ean delete Deleting directory C Users warden Subversion P1aSMA_Scenarios scenarios scenario_b4_common build uninstall delete Deleting directory C Users warden Subversion P1aSMA_Core_Vani11a app scenarios
6. lt attribute key operatesFor value http plasma informatik uni bremen de owl default_northrange owl TZI_Logistics gt lt attribute key representsObject value http plasma informatik uni bremen de owl default_northrange owl Truck TZl_Logistics 03 gt lt agent gt lt mappings gt 4 2 3 World Model lt worldmodel gt optional once Provides some world model related settings The creation of the world model itself is explained in Section 5 4 ontology optional Determines if run time changes of the world model e g adding objects should be fully backed up by the ontology true or if a slightly simpler data structure should be used false The default value is false which is also recommended for performance reasons 2For this effect see Section 5 5 4 2 DOMAIN CONFIGURATION 25 lt database gt optional once Provides informations to connect to a database representing the transport infrastructure This is only required if the infrastructure will be imported from that database see Sec tion 5 4 4 instead of provided directly through the ontology see Section 5 4 1 url required The URL to the database This might require a jdbc postresql scheme e g jdbc postgresql localhost 5432 postgis user required The database user password required The password of that user enabled optional Setting this attribute to false disables the usage of the database infras tructure import without re
7. IMPLEMENTATION pgadmin and creating a new database and if desired a new user that should contain the imported data PlaSMA provides a tool named plasma datamimport osm to convert osm files into a format that can be handled by PlaSMA You can start it by executing the provided script for your OS in the PlaSMA installation directory Windows prompt gt osm import bat Linux prompt gt osm import sh Connecting to the Database Once you have created the database and a user to store the data from an osm file you can use the PlaSMA OpenStreetMap Importer to create the needed tables and store the provided data You can start the Importer by executing the osm import sh or osm import bat in the plasma main directory As soon as you has started the program a dialog appears 5 7 that prompts you to insert all needed data to connect to postgres database created in the previous step Configure Database Y YD x Server localhost 5432 Database postgis Username postgres Password 1 Connect Cancel Figure 5 7 How to connect to an existing Database If you provided the correct data a connection to the database will be established and the dialog will close and the main window will appear After the first start the PlaSMA OpenStreetMap Importer will store all provided information excluded the provided password Main Interface The main window 5 8 can be
8. a simulation scenario but rather want to visualize it for demonstration purposes you might skip the installation of PostgreSQL Some scenarios might need a PostgreSQL PostGIS installation the scenario maintainer can tell you if an installation is required An installation at a later point is also still possible If you choose to skip the installation of PostrgreSQL note that you have to configure the PlaSMA server accordingly see Section 2 2 2 If you don t intend to actually implement own agents i e writing Java code the JRE is fine for you 3You can check Java version by typing java version on command line The internal version number for Java7 or 6 will be indicated as 1 7 or 1 6 though 4The procedure is explained here http support microsoft com default aspx scid kb en us 310519 5If you want to install Java from Oracle you can do so by downloading the linux specific archive from the the Oracle download page http www oracle com technetwork java javase downloads index htm1 and follow the instructions from Oracle 2 1 INSTALLATION AND SETUP OF REQUIRED AND OPTIONAL SOFTWARE 5 Installation procedure Windows Before installing PostgreSQL please ensure that you have removed all files from previous PostgreSQL installations When removing previous installation files you might be prompted to restart your Windows operating system It is recommended that most users download the binary distribution for Windows Latest ver
9. actions might depend on situation requirements In any case transitions have to be deterministic The concept of Finite State Machines can be easily adopted to the design of a complex behaviour You will often find behaviours that directly depend on the results of other behaviours For instance the 101f more than one transition leads from one state to a following every transition needs a unique condition to determine the correct following state 48 CHAPTER 5 SCENARIO IMPLEMENTATION driving behaviour of a truck agent might depend on the result of a routing behaviour Instead of mixing them up into one oversized behaviour or exchanging wake up calls between them it might be more accurate to combine these behaviours to a Finite State Machine This way all behaviours can be developed and implemented without knowledge about each other It is easier to add remove or swap behaviours and allows to re use behaviours from other agents without making major changes to the source code Implementation of a FSM Behaviour The idea is simple Every state of the FSM represents a certain behaviour Whenever a behaviour terminates a result value will be returned when its onEnd method is called The result can be used as a transition condition to determine which behaviour ought to be scheduled next To keep this result readable and traceable it may be useful to define a Java enum type for each behaviour that contains all possible results For i
10. agent of the simulation core causes an error or if an unpredicted error occurs in the scenario this will be reported and processed at a central location In each case this will lead to an abort of the simulation and the error will be forwarded to the graphical user interface which will show a window with the cause of the error Depending on the configuration of the simulation core see Section 4 3 an error case will trigger a backtrace on each connected simulation computer Detailed protocol messages of the agents promptly before the occurrence of the error will be logged They will appear depending on the configuration either in the database or on the terminal on which the simulation was started This mechanism is based on a huge amount of protocol messages which are buffered by the system although they would be actually discarded by the current log level This is required because most errors of the simulation are caused by concurrency That means that they would appear infrequently or never if the simulation is executed with a debugger or at another log level because of the slow down of the execution The buffering of the detailed protocol messages does not lead to any performance losses but makes the search for errors much easier Scenario Configuration Simulations runs their visualizations and the logging of data of the scenario activities are configured in a scenario specific XML file The format of this file is specified by the XSD schem
11. divided in the 4 components described below e Menu The menu at the upper top of the window contains additional functionality like stor ing loading configuration Information about loading and storing data can be found in paragraph Saving Loading data e Osmosis Import This panel on the left hand site of the window contains some options for importing osm data See Importing OSM Data for detailed instructions e Filter This panel on the upper right of the window contains settings about various filter configu rations Theses settings are described in detail in the paragraph Filtering relevant data e Log The Log panel on the lower right of the window contains information about the progress of pending operations All errors warnings and verbose informations can be found here 6See chapter 2 1 1 about seeting up the database for more information 5 4 WORLD MODEL DESIGN 41 Yu Plasma OpenStreetMap Import VOY File About Osmosis Import Filter Select the relevant types of the Press the start button to reduce the complexity of the overall graph without infrastructure and determine a effects on the granularity of the infrastructure model OSM file that will be imported Klick the start button to start the import process Option 1 Save the graph infrastructure as PlaSMA compatible ontology C Export infrastructure to ontology J Relevant Edge Types Ontology File ls German Autobahn Option 2 Include public
12. likely the core of each Action It determines the necessary changes to the world model and stores them into an ObjectModelDelta In this case after the scheduled loading time is passed the appropriate changes are provided to indicate that the load is now stored in the storageFacility If the required time hasn t passed yet nothing changed and an empty delta is returned Override public WorldModelChanges execute final long currentTime final ObjectModelDelta delta new ObjectModelDelta if currentTime gt finishingTime delta addChangedObjectPosition load goalPosition delta addPhysicalObjectObjectProperty storageFacility ObjectModel HAS_LOADED_PROPERTY_ID load this complt true return delta During the load process the load s position is set to a transitional position since it s not placed in its old position anymore but also not stored in the storage facility yet Override public WorldModelChanges transitionState final long currentTime final ObjectModelDelta delta new ObjectModelDelta final ObjectPosition pos this getWorldModel getPositionByID load if pos instanceof DefinitePosition this abort Illegal initial position type pos getType else this currentPosition new TransitionalObjectPosition DefinitePosition pos goalPosition delta addChangedObjectPosition load currentPosition return delta The needsExecution method may indicate if th
13. same directory as postgresql conf 10 Technically interested users can find the configuration file as a XSD schema at lt plasma_dir gt etc serverconfig xsd 2 2 PLASMA SERVER CONFIGURATION 11 lt scenariodir gt optional multiple Describes additional directories where the PlaSMA server searches for scenarios The default scenario directory lt plasma_dir gt scenarios is included automatically without being configured explicitly The path can be specified in lt scenariodir gt s path attribute path required Sets the path for additional scenarios lt logging gt required once Configures the way PlaSMA server emits messages and where these are stored or displayed The attributes of the lt ogging gt element are loglevel required lt sets the standard log level according to the implemented log level types including NONE ERROR WARNING INFO DEBUG and TRACE Generally there are two output channels for the server the database and the console output Consequently they are configured first in the lt logging gt element lt database gt optional once This element configures the database connection of the server If omitted the database is turned off as well as if the enabled attribute is set to false The database settings should be configured according to the database setup see also Section 2 1 1 lt database gt contains these attributes enabled optional If false the database connection is turned off and there is no
14. scenario_b4_common icompile mkdir Created dir C Users warden Subversion P1laSMA_Scenarios scenarios scenario_b4_common bui 1d 5 javac Compiling 36 source files to C Users warden Subversion P1aSMA_Scenarios scenarios scenario_b4_common bui 1d jar _ jar Building jar C Users warden Subversion P1aSMA_Scenarios scenarios scenario_b4_common bui 1d scenario_b4_com jmon jar jdeploy jar mkdir Created dir C Users warden Subversion P1aSMA_Core_Vani1la app scenarios scenario_b4_common copy Copying 1 file to C Users warden Subversion P1aSMA_Core_Vani 1 la app scenarios scenario_b4_common jdeploy resources E A copy Copying 2 files to C Users warden Subversion P1aSMA_Core_Vani 11a app scenarios scenario_b4_common deploy BUILD SUCCESSFUL Total time 5 seconds x Figure 5 4 An example for an Ant build session directory compile Compilation of scenario s Java source files jar Packaging of scenario classes into a Java archive jar deploy Deploys the scenario with all its resources to the PlaSMA scenario directory Thereby the scenario will be available for execution in PlaSMA uninstall Uninstalls a scenario by deleting the corresponding sub directory in the PlaSMA directory for deployed scenarios 5 3 1 Using the Ant Based Build System from Eclipse In the following we assume that the scenario developer has already created a new Java project for the new scenario Thus when starting from the scenar
15. transport information Select a folder to create an OWL D motorway ontology file for saving the physical stops e g bus stops and the itinerary i information O E motorway_link fs Federal Road GDS trunk Timetable Directory J DB trunk link y primary C Import from X10 files D primary link 10 Directory J 0 secondary C Include public transport infrastructure and timetables gt Option 4 Select the OSM ID normaly a 9 digit number of a node that is definetely a GD secondary link part of the main graph All nodes that cannot be reached from the selected node will Inner City Road be excluded from the result Example 123456789 dia Railway Node ID E Public Transport start Log Information Connecting to jdbc postgresd localhost 5432 plasma Connected to database OSM File search start Figure 5 8 OSM Importer main interface Importing OSM Data All settings for importing osm data into a plasma compatible format can be found in the Osmosis Import panel To import data to the database you created earlier you have to follow the next steps e Specify all edge types you wish to import from the tree on the left Only selected data types will be imported e Choose the osm file from your file system you want to import into the database You can choose a file by clicking on the search button and browse y
16. ION USER CONFIGURATION USER CONFIGURATION ie Fa PlaSMA Core Build Management Location of the PlaSMA Core System El PISMA 34 Conania Codebese 4 E5 PlaSMA Scenario Germany For Windows users please note that in the specification of the path to the E BROMA Soenen Northern Germany PlaSMA installation directory single backslashes must be substituted E PlaSMA Scenario Template with double backslashes as in 18 clean eon shared plasma location C Users myUser Plasma compile from shared a deploy default from shared Furthermore please specify an absolute rather than a relative path 18 deploy jar from shared plasma location lt path to plasma installation gt to deploy resources from shared 18 jar from shared If the scenario depends on jars from other scenarios the latter need to to pathTest from shared be deployed properly to your Plasma installation Afterwards you can 18 uninstall from shared configure the build process such that the deployed jars from the scenarios D PlaSMA Scenario Test PlaSMA Worldmodel you depend on are used during compilation You do this by uncommenting the following line in case of dependencies scenario hasDependencies STATIC CONFIGURATION STATIC CONFIGURATION STATIC CONFIGURATION STATIC CONFI PlaSMA Core Libraries Required to build the Scenario Pplasma libs S plasma location lib Scenario Build Structure acenario sre sre 4 m r
17. If you do not use the additional option import from X10 file the importer will auto generate timetables for you In this case for each generated timetable each bus tram starts at every full hour and will need a two minutes ride per stop and wait on each stop for one minute It is also possible to use X10 data to specify timetables and get a better public transport infrastructure Currently you can only simulate two different types of vehicles called buses and trams and you have to clone one week day for every day of the week To use this feature you have to specify the X10 directory and do some manual manipulation of the PlaSMA OpenStreetMap Importer Properties file See Configuration of the PlaSMA OpenStreetMap Importer for more information e Remove irrelevant sub graphs You can also remove subgraphs that are not connected to the graph you are interested in To do so you have to specify a node ID a 9 digit number corresponding to on node in the graph you want to keep All nodes that can not be reached from that node will get removed from the graph Saving Loading data The PlaSMA OpenStreetMap Importer loads the last used configuration as default when starting the program You can also save the current data for later use or to share configuration with other users To do so you can simply click on the Save Setting or Load Setting Entry in the File menu This will open a file chooser Dialog where you can select the file you want to save load By def
18. MAIN CONFIGURATION 23 log a warning message The GUI client will show a corresponding warning and prompt to proceed or to stop simulation A timeout event might indicate that an agent got stuck This attribute uses the same format as the maxSimLength attribute see Section 4 1 4 2 Domain Configuration A simulation scenario domain is primarily defined by the agents to be executed These simulation agents are parallel logical simulation processes Each simulation agent may represent a set of physical objects or may provide some service not directly related to a specific physical object This section describes how to configure the agents to be executed and optional settings for the visualization of physical objects they handle lt domain gt required once The embracing XML tag for the scenario domain configuration The following sections describe its subelements 4 2 1 Scenario Ontology Specification lt scenario gt required once Defines the scenario The content of the XML tag is the scenario OWL ontology URI Uniform Resource Identifier This ontology URI specifies the scenario s traffic network as well as all simulated objects and agents with their relations and attributes The ontology is denoted by an logical URI PlaSMA resolves this URI to a physical URL In general the scenario will be located in the lt scenario gt owl sub directory or in the general PlaSMA ontology directory etc ow1 For instance the URL owl scenario ow
19. One can ensure to receive only appropriate messages with the help of filters MessageTemplate It is also possible to put message back in the inbox if it turns out that they are out of place see Agent putBack ACLMessage 5 5 3 Agent Behaviours The behaviour of the agents is realized through jade core behaviours Behaviour class and its subclasses in the same package When subclassing from these Behaviour classes in turn one has to override the action method To find out more about JADE behaviours you are referred to the JADE 2 There are changes for future versions PlaSMA in preparation which also allow blocking wait The specified waiting period would be considered simulation time and the agent would automatically end its cycle if no matching message is available 5 5 AGENT IMPLEMENTATION 47 manual BCTRO7 pp 24 It is possible to have several distinct types of JADE behaviours mixed up for one agent so that both sequential and parallel behaviours are being processed Up to now the parallel behaviours are not fully supported in PlaSMA In general behaviors of simulation agents should be synchronized This is achieved by the implementation of the Java interface UserBehaviour from within one s custom behaviour class Instances of this behaviour should then be added to the agent with the addUserBehaviour method In extremely rare cases it can also be useful to use unsynchronized behaviours Those can still be added with the ba
20. ab shows the Server Log WorldWind Visualization see Figure 3 2 is based on NASA World Wind an OpenGL based 3D engine for geographical visualizations Thus scenarios are expected to be located somewhere on the Earth surface The menu items WorldWind Layers and PlaSMA Layers allows you to select various layers displayed on that surface The first category includes layers for e g satellite images road maps place names and country borders The real world data for these layers is retrieved from the Internet therefore you need an Internet connection Anyhow World Wind will cache loaded images that will be displayed even without Internet connectivity The latter category provides visualization of physical objects and the traffic infrastructure in the respective scenario Note Besides these default scenario layers available visualizations will depend on the configuration of the scenario see Section 4 2 The setup of your scenario can be viewed and edited in the Scenario Configuration tab It contains a wide variety of options from modifying basic scenario parameters see Figure 3 3 to add or remove scenario agents You may change the displayed settings according to your simulation preference and purpose e g number of runs and length of each run All modifications will be applied automatically as soon as you start the scenario run You may save the modified configuration for later usage by using the buttons on top of the tab Amongst o
21. ables from scratch Explain Data Output Messages History ready Unix Ln 1 Coli Chi Figure 2 3 Loading database schema via pgAdmin III Optional Setting up network access The following configuration instructions refer to distributed computation of your simulation In case you would not like to have computation distributed within your network you can step over the following configuration You can edit the network setting via Tools gt Server Configuration pg_hba conf In this configuration file you have to add all hosts of you network which in turn may access the PostgreSQL database on your local machine The following format should be applied for IP addresses lt P Adresse gt 32 2 1 INSTALLATION AND SETUP OF REQUIRED AND OPTIONAL SOFTWARE 9 Base Configuration Method 3 Command Line In this subsection we will guide you through PostgreSQL database setup using the command line tool First open a command line window On Windows change to the bin directory of PostgreSQL You will find this directory as a subdirectory within the PostgreSQL installation path e g C Program Files PostgreSQL 9 1 bin Now we will create a new user role for our PlaSMA database with permissions to create other users and databases and with an encrypted password Enter and execute the following command You may choose an arbitrary user name Windows prompt gt createuser exe U postgres s P E lt user_
22. and processed if necessary For example an action could move a truck from A to B over a road stretch edge in the modelled traffic infrastructure Depending on the simulation time that passes between the creation and the following synchronization step the truck will be shifted on the edge towards B If the truck has not yet reached B the action will not be marked as complete and will be re scheduled automatically at the next desired synchronization step In order to create your own action it is necessary to extend the abstract class Action For better understanding the following text gives a short description of the important methods e WorldModelQuery getWorldModel Provides read only access to the world model after action registration The action should use this method to query needed information on the current state of the world model e long getCreationTime Returns the simulation time as UNIX millisecond timestamp when the action was created The creation time corresponds to the time it was registered with the world model e void init This method initialized the action as intended by the action programmer It is called by the world model upon registration In contrast to the constructor the action creation time and the world model will be accessible when the method is called e boolean checkPreconditions throws ActionException This method will be called by the world model after registration and initialization of the action If t
23. anning modes and operates for a specific company Its configuration contains the following attributes lt agent description TZI Logistics Forwarding Agency name Management TZl_Logistics class org tzi plasma toolkit forwardingagency agent Agent_ForwardingAgency gt lt attribute key run_mode value CENTRAL_PLANNING gt lt attribute key operatesFor value http plasma informatik uni bremen de owl default_northrange owl TZI_Logistics gt lt agent gt Recommended Attribute Access To access those attributes it is recommended to implement the ConfigKeyInterface as an enum Given that enum the attribute can be accessed through the agent s getParameter ConfigKeyInterface Class lt T gt method or getParameterList ConfigkeyInterface Class lt T gt if the attribute con tains multiple values The first parameter is the corresponding enum entry of the attribute s key spelling needs to be exactly the same and the second parameter determines the type of the attribute s value to avoid the usually required cast assuming a CustomAttribute Enum String mode getParameter CustomAttribute run_mode String class The previously mentioned predefined keys are provided through the Ontology0bjProperty enum Since ontological references are always represented as BasicRIs the agent has special methods to skip that second type parameter namely getReferenced0bjFor Ontology0bjProperty and getRefere
24. ario gt BB Server Log E WoridWind Visualisation 9 Scenario Configuration Bj Save Save As Scenario Options Agent Options Agents T locationdanager Basic Options Ontological References Y a Management SFB_Logistics Agent Name TruckOp TZI Logistics 01 representsObject Zi Y la Management TZL Logistics Java Class org tziplasma toolkittruckagent Agent_TruckMa y E TransshippingFacility 0 5 7 lg TruckOp TZI Logistics 01 StorageFacility 1 78 7 MB TruckOp TZL Logistics 02 Log Levet NONE x E ISOContainerStorage 0 61 EIA TruckOp TZ1 Logistics03 Y Creation Date 01 01 2009 09 00 00 E StaticStorageFacility 0 61 VI TruckOp TZI_Logistics 04 a El amp Truck 1 16 TruckOp TZL Logistics 05 Description TZI Truck Management Agent 1 E amp TruckSFB_Logistics 04 TruckOp TZL Logistics 06 E amp Truck SFB_Logistics 05 dla TruckOp TZI_Logistics 07 Atihutes E amp Truck TZL Logistics 08 ba TruckOp TZI Logistics 08 Key Value E amp Truck SFB_Logistics 02 7 lla TruckOp SFB_Logistics 01 E amp Truck TZL Logistics 07 E Y lg TruckOp SFB_Logistics 02 E amp Truck SFB_Logistics 03 Y TruckOp SFB_Logistics 03 E amp Truck TZL Logistics 06 M TruckOp SFB_Logistics 04 EJ amp Truck TZL Logistics 05 TruckOp SFB_ Logistics E amp Truck SFB_Logistics 01 WA TruckOp SFB_ Logi E amp Truck TZLLogistics 04 Ty TruckOp SF8 Logistics E amp Truck TZL Logistics 03 I Truc
25. articular system model that was subject to simulation For further details about the logging and computational metric we encourage you to continue reading in Section 4 3 and Section 5 5 10 Generally there are three different ways to configure PostgreSQL database for using PlaSMA One of these implies to use the pgAdmin III tool the other two ways make use of a command line tool We recommend to use the ready to use command line tool together with a configuration script as described in the following subsection Note that in order to have your simulation runs being computed in a distributed fashion you will most probably have to make additional settings as described in Section 2 1 1 For all methods you should ensure that the PostgreSQL service has started and is running successfully Windows You can check the status of the PostgreSQL Windows service in the Windows services overview via the system panel Ubuntu You can checkt the status of PostgreSQL by typing prompt gt etc init d postgresql status If the service is not running type prompt gt sudo etc init d postgresql start Base Configuration Method 1 Initialization Script As a first option you can use the initdb script file provided in lt plasma_dir gt that can create both the PlaSMA database user and the PlaSMA database and then initialize the database automatically without any user interaction except entering a password The script can be used as
26. ask at hand where the focus is primarily on an ontology based modeling of this world model Section 5 5 is concerned with the development of new types of agents The chapter is concluded with an introduction to simulation time synchronization and the mechanism used in PlaSMA PlaSMA Setup The following section describes how to set up PlaSMA for the major branches of operating systems the development team is actively supporting namely Linux and Microsoft Windows 7 In the following OS specific setup instructions we assume a standard installation of the respective OS Wherever special software packages are required for successful deployment of the PlaSMA multi agent based simulation environment this is explicitly noted The PlaSMA simulation environment mainly consists of four software components e The simulation core PlaSMA Core that controls the execution of simulation scenarios and can be thought of as an intermediate level built on top of the functionality provided by the underlying FIPA compliant multi agent system JADE e A optional database system that stores all the data and performance metrics generated by a running scenario for analysis purposes e The PlaSMA Client for selection and control of simulation scenarios by the user and for their visualization if required e The actual agents that represent logistic objects and populate the scenarios The OS specific release version is packaged as tar bz2 or Zip archive and is availab
27. atabase objects E Can create roles Y Can modify catalog directly Comment E Can initiate streaming replication and backups Use Slony Help CE cancel Help Cancel Help Cancel Figure 2 1 Creating a new login role in pgAdmin III Creating a new database In the next step we will create a database dedicated to PlaSMA This can be done via a context menu over the Object Browser list view as well Open a context menu over the list entry Databases Then choose New Database Again a new application dialog comes up in which you can specify the new database according to the screenshots in Figure 2 2 It is important to set access permissions for the previously created login role Setting this user as the owner of your dedicated database is just fine PlaSMA will assume that you use UTF 8 character encoding in this database which is set by default When you confirm these setting a new database will be created 8 CHAPTER 2 PLASMA SETUP New Database New Database a Properties Definition variables Privileges soL Properties Definition variables Privileges 5QL Name plasma Encoding urre om Template x Owner plasma Tablespace defaut tablespace gt x Collation y Character type Connection Limit 1 Comment Schema restriction C i a Ged Figure 2 2 Creating a new database in pgAdmin III Setting up the database After creation of the databas
28. atter is taken into account 5 5 10 Performance Metric Logging In addition to the message based logging system there is the possibility of using typed performance metrics called performance indicators in PlaSMA Performance indicators can be used to log agent metrics to the database such as driven kilometers speed truck utilisation profits etc and analyze them when the simulation is done Usage Performance indicators are identified by keys Such a key contains some basic informations about the performance indicator e A unique name identifier for the performance indicator e A short description of the performance indicator e The unit of the values the performance indicator will store If only a small number of performance indicators is used each can be created separately using the PerformanceIndicatorKey class If a lot of indicators are used it is recommended to store them in an Enum implementing the PIKeysInterface interface Examples can be found in the toolkit source code by searching for classes that implement the PIKeysInterface interface 5 5 AGENT IMPLEMENTATION 55 PlKeysInterface computationTimelndicator new PerformancelndicatorKey sp_cmp_time Shortest path computation time ms To access a performance indicator with the previously created key an instance of the PILogger is required Such an instance can be easily created using the PILogger forOwner SimulationAgent method The simulati
29. ault the files are stored with as iosm file These files are readable in any common text editor and can also be changed by hand if desired See Configuration of the PlaSMA OpenStreetMap Importer for more details Configuration of the PlaSMA OpenStreetMap Importer There are two different configuration files for the PlaSMA OpenStreetMap Importer Both files are build of key value pairs separated by a sign The part before the equality sign is the key that is used by the importer that is used to find the corresponding value You should always only change the values of these pairs otherwise the program might not work as expected If the Importer can t find one of the files at startup a file filed with default values and blank fields will created automatically Database Settings The database settings are stored after every successfully established connection There should be no need to specify any data here by hand but if wished you can do so Be aware of that no password data will be saved for database access e DEFAULT_SERVER The Server IP you want to connect to default localhost e DEFAULT_PORT The Port on which the server listenes to incomming database connections default 5432 e DEFAULT_DB The name of the database you want to access e DEFAULT_USER The name of the user with the needed rights to access the database Filter Settings Filter Files are used to save the settings of the PlaSMA OpenStreetMap Importer There are also some small add
30. chuldt and S Werner Quality Criteria for Multiagent Based Simulations with Conservative Synchronisation In 13th ASIM Dedicated Conference on Simulation in Production and Logistics ASIM 2008 pages 177 186 Berlin Germany 2008 Fraunhofer IRB Verlag D R Jefferson Virtual Time Il Storage Management in Conservative and Optimistic Systems In Proceedings of the Ninth Annual ACM Symposium on Principles of Distributed Computing PODC 1990 pages 75 89 Quebec Canada August 22 24 1990 ACM Press P F Patel Schneider P Hayes and I Horrocks OWL Web Ontology Language Semantics and Abstract Syntax Available from http www w3 org TR owl semantics February 10 2004 H V D Parunak R Savit and R L Riolo Agent Based Modeling vs Equation Based Modeling A Case Study and Users Guide In Multi Agent Systems and Agent Based Simulation First International Workshop MABS 1998 volume 1534 of Lecture Notes in Computer Science pages 10 25 Paris France July 4 6 1998 Springer A Schuldt J D Gehrke and S Werner Designing a Simulation Middleware for FIPA Multiagent Systems In 2008 IEEE WIC ACM International Conference on Web Intelligence and Intelligent Agent Technology WI IAT 2008 pages 109 113 Sydney Australia 2008 IEEE Computer Society Press 59
31. ck is conducted that resets the agent to its former state at LVT t However optimistic synchronization is more complicated in implementation and has potentially high requirements regarding space Fuj00 Conducting a rollback might require re turning many steps back in time Hence all preceding states of every agent must be stored By contrast conservative synchronization prevents causality problems by ensuring the correct order of event processing at each time This in turn restricts the speedup achievable by parallelism There are numerous approaches to conservative synchronization which are described in Fuj00 The synchronization mechanism is an important choice when implementing a MABS system 5 6 2 Synchronization in PlaSMA The PlaSMA system applies coordinated conservative synchronization with a simulation controller hierarchy As a consequence all active simulation agents have the same LVT resulting in a global virtual time This approach applied by PlaSMA constrains speedup because agents may be forced to wait for other agents One should set the scenario minSyncDist property not too small in order to avoid such effects Better use values greater or equal 20 seconds Details about the synchronisation implementation are documented in SGW08 23See Section 4 1 1 BCTRO7 BvHH 04 Dav00 Fuj00 Geh08 GOBO7 GS09 GSwos Jef90 PSHH04 PSR98 SGWOo8 Bibliography Fabio Bellifem
32. ction POINT geom AS POINT the_geom AS L bbox AS LINES linestring AS LI D bbox AS POLY bbox AS LINES linestring AS LI ways filtered __LINESTRING E bli E POLYGON __bbox AS POLY nn nn A Figure 5 11 Select the data that you want to visualize 5 5 AGENT IMPLEMENTATION 45 lt worldmodel gt lt database enabled true url jdbc postgresql localhost 5432 lt database name gt user postgres password xxx gt Note If you do not use an ontology representation for the transport infrastructure would recommend to set the positions of physical objects in your scenario owl in the following way lt tlo positionedAt gt lt tlo Coordinates rdf about name gt lt tlo latitude rdf datatype amp xsd float gt 53 08766 lt tlo latitude gt lt tlo longitude rdf datatype amp xsd float gt 8 722644 lt tlo longitude gt lt tlo Coordinates gt lt tlo positionedAt gt It is inconvenient to lock for a specific node in the database If there do not exist a node on a specified coordinate a new node is created and connected to the nearest existing node The length of the new edge is the linear distance times 1 3 5 5 Agent Implementation For simulation purposes real world objects and entities as well as environmental processes can be represented by parallel logical simulation processes called simulation agents A simulation run in PlaSMA is a forward directed computa
33. d filter them to get a clear arrangement Furthermore you can define a special log level for each category This allows you to log messages from a specific category more or less extensive or even not at all If a category is specified for a logging message the log level of the category will decide about the actual logging of the messages and not anymore the overall log level The categories are organized hierarchically and are identified by the path The respective levels are divided by a dot e g System Communication Input Thus the categories are organized in a tree structure This means that a category inherits the log level of the super class if it does not specify its own log level If no log level is specified for a top level category the global default log level is used There exist some predefined categories in the Java class LogCategory as static variables whereas System shall be used for system messages only which is thus reserved for the simulation core Note that category identifiers are handled case sensitive 4 3 3 Agent specific Logging It is possible to assign a specific log level to each agent that overrides the default scenario log level This log level can be set at the respective lt agent gt entry in the XML configuration see Section 4 2 2 4 3 4 XML Configuration lt logging gt optional once Sets the scenario s default log level and encloses the specific log category definitions loglevel required The
34. d it convenient to reuse parts of the classes developed in one base scenario in another dependent scenario The Ant based build process accommodates this particular scenario in the following way If you configure a dependent scenario you can uncomment the flag scenario hasDependencies in the configuration file Then during compilation Ant makes sure the Jar archives of previously deployed PlaSMA scenarios are present in the classpath during compilation scenario_example src lib optional owl scenario owl config images optional build xml configuration properties Figure 5 1 Schema of the structure of a typical PlaSMA scenario 5 3 SCENARIO BUILD PROCESS 31 Once the configuration properties file is prepared the Ant based build system is ready for use Note that if you try to use Ant without the configuration being present the build process will abort with an error message asking you to provide said file first 5 3 Scenario Build Process The scenario build process is based on the Apache Ant tool A standard build xml file is part of each scenario which supports the following set of commands called targets clean Cleans the scenario workbench by deleting all compiled Java class files from the scenario build buildxm Ant X N EE Outline E Task List Z0 SCENARIO USER CONFIGURATION LP QORK PlaSMA Core Application Launcher USER CONFIGRUATION USER CONFIGURAT
35. e Note that this waiting period refers to the real not the simulation time In simulation the non blocking as well as the blocking method will return null if no message has been received at the beginning of the current agent processing cycle The simulation inbox management will intercept all later messages to insure simulation run causality and reproducibility cf GSW08 When retrieving messages note the following e Using the blocking methods for retrieving messages from the inbox stops all other JADE behaviours of the agent until a message could be received or the waiting period expired Moreover the original JADE blocking receive methods with no timeout could yield into a deadlock in simulation because there is no guarantee that the desired message will be available in the current cycle Thus SimulationAgent always uses a default timeout of 5 seconds e Querying the inbox with waiting time is generally to be avoided as hereby simulation will be unnecessarily delayed Such a method call should rather be repeated in a following cycle with other simulation time It should also be noted that the waiting time is real time not simulation time e lf the receive method returns a message the message is being dequeued from an internal message queue of the JADE agent Then this message will be no longer available in the inbox Therefore a single behaviour should not remove a message from the inbox that is regarded to another behaviour
36. e we load a database schema which describes the structure of the PlaSMA database The database structure is generated by the SQL script found in lt plasma_dir gt etc sql plasma_datamodel sql This script contains SQL statements which create tables fields and indices needed In order to execute the script select the newly created database in the Object Browser and choose from the menu bar Tools gt Query tool A new application dialog comes up within you can load the script file and execute it see Figure 2 3 Choose Query gt Execute from the menu bar to run the script once Thereafter the PlaSMA database is ready for use gt Query plasma on postgres localhost 5432 E plasma plasma_core app etc sql plasma_datamodel sql File Edit Query Favourites Macros View Help B a o Af D be da Ta a Q O plasma on postgres localhost 5432 Scratch pad SQL Editor Graphical Query Builder Previous queries Zi Delete Delete All drop old tables a m DROP TABLE IF EXISTS log messages DROP TABLE IF EXISTS communications DROP TABLE IF EXISTS keymeasures_complex DROP TABLE IF EXISTS keymeasures_ntuple DROP TABLE IF EXISTS keymeasures_double DROP TABLE IF EXISTS keymeasures_long DROP TABLE IF EXISTS keymeasures_string DROP TABLE IF EXISTS keymeasures DROP TABLE IF EXISTS sys_messages DROP TABLE IF EXISTS worldevents DROP TABLE IF EXISTS server_state create new t
37. e Action needs re evaluation due to the latest changes to the world model which is not the case in this example Override public boolean needsExecution final WorldModelChanges changes return false J 52 CHAPTER 5 SCENARIO IMPLEMENTATION The last method nextExecutionTime provides the simulation timestamp at which the Action should be executed the next time In this example the next execution time is directly after the loading process has finished Override public long nextExecutionTime return finishing Time For more examples take a look into the toolkit source code which can be found in the toolkit_src directory Common actions are located in toolkit_src org tzi plasma worldmodel actions and toolkit_src org tzi plasma toolkit load agent actions The code of this example can also be found in the toolkit source in toolkit_src org tzi plasma worldmodel actions LoadAction java 5 5 7 Agent Attributes To be able to configure an agent for different scenarios it is possible to provide arguments in the scenario configuration that the agent can access at run time This is also a convenient way to link an agent to ontological instances PlaSMA provides three predefined attribute keys for common ontology references representsObject represents nfrastructure and operatesFor As an example take a look at the Agent_ForwardingAgency This class represents a forwarding agency that supports multiple pl
38. e config xsd in the directory lt plasma_dir gt etc For execution all scenarios have to be deployed to a subdirectory in lt plasma_dir gt scenarios in whose subdirectory config the XML configuration file is saved Besides manually editing the configuration file with a text editor the scenario configuration can also be edited with the visual client see Section 3 2 1 If the desired scenario was selected from the Select Scenario menu the configuration can be modified in the Scenario Configuration tab A scenario template without any agents can be found in the lt plasma_dir gt scenarios scenario_template config directory and should be available in the Select Scenario menu as PlaSMA Scenario Template The following describes the creation and modification of a scenario configuration XML file 4 1 General Simulation Settings lt simulation gt required once ls the top element of simulation run description and contains the following attributes name required Defines the name of the simulation scenario This name is shown during scenario selection in the client program repeats optional Gives the number of iterations of the simulation run If this attribute is missing l is used as default value This means that the simulation runs only once While starting more then one simulation run the simulation time progress continuously in subsequent runs maxSimLength required Specifies the simulation time after which the simulation should be sto
39. el of detail of each messages e g errors get the level ERROR Depending on logging detail configuration a generated message will be actually logged or ignored The log level declared in the scenario s XML configuration is the determination base A message with the same or a log level above the specified level will be logged Otherwise it is discarded The lowest log level is the level TRACE whose messages are only logged if the log level TRACE is specified For instance if the log level WARNING is specified only messages with the log level WARNING and ERROR are logged Here is an overview of the log level hierarchy in increasing order of detail 1 NONE can only by specified in the configuration file With this level all messages are discarded 2 ERROR is the highest log level for messages and should be used to log an error that usually causes a simulation run to fail 3 WARNING denotes a warning that does not cause the simulation to fail necessarily 4 INFO Messages that are helpful to keep track of important events in the scenario but not generated too frequent 5 DEBUG for analysis of errors and problems in the programming of agents or their interaction 6 TRACE is the lowest log level and should be used to trace the exact behaviour of an agent The simulation will be slowed down by a high amount of these messages 4 3 SIMULATION LOGGING 27 4 3 2 Logging Categories Categories provide the ability to group messages an
40. en de owl tlo owl gt lt ENTITY test http plasma informatik uni bremen de owl test owl gt lt rdf RDF xmlns http plasma informatik uni bremen de owl test owl xml base http plasma informatik uni bremen de owl test owl xmlns rdfs http www w3 org 2000 01 rdf schema xmlns owl11 http www w3 org 2006 12 owl11 xmlns tlo http plasma informatik uni bremen de owl tlo owl xmlns 0w111xm1 http ww w3 org 2006 12 0w111 xm1 xmlns test http plasma informatik uni bremen de owl test owl xmlns owl http www w3 org 2002 07 owl xmlns xsd http www w3 org 2001 XMLSchema xmlns rdf http www w3 org 1999 02 22 rdf syntax ns gt lt owl Ontology rdf about gt lt owl imports rdf resource http plasma informatik uni bremen de owl tlo owl gt lt owl Ontology gt lt Nodes locations gt lt tlo Vertex rdf about Bremen gt lt tlo latitude rdf datatype amp xsd float gt 53 0759 lt tlo latitude gt lt tlo longitude rdf datatype amp xsd float gt 8 8073 lt tlo longitude gt lt tlo Vertex gt lt tlo Vertex rdf about Hamburg gt lt tlo longitude rdf datatype xsd float gt 53 5506 lt tlo longitude gt lt tlo latitude rdf datatype amp xsd float gt 9 9933 lt tlo latitude gt lt tlo Vertex gt lt Edges roads gt lt tlo Edge rdf about Bremen_Hamburg gt lt tlo startingAt rdf resource Bremen gt lt tlo directedTo rdf resource Hamburg gt
41. es A A eG 27 4 3 3 Agent specific Logging 2 oaa a ee 27 4 3 4 XML Configuration aaa a 27 5 Scenario Implementation 5 1 Overview of the General Scenario Structure 5 2 Scenario Build Configuration 5 3 Scenario Build Process 5 3 1 5 3 2 5 4 1 5 4 2 5 4 3 5 4 4 5 5 Agent Implementation 551 5 5 2 5 5 3 5 5 4 5 55 5 5 6 5 5 7 5 5 8 5 5 9 5 5 10 55 11 5 6 Time Management and Synchronization 5 6 1 5 0 2 Bibliography Using the Ant Based Build System from Eclipse Using the Ant Based Build System from the Command Line 5 4 World Model Design Physical World Infrastructure o o World Model and Agents o e e Scenario Ontologies o o o e eee How to Import a Transport Infrastructure from OpenStreetMap Agent Setup se co a Boe a a a a Reception and Processing of ACL Messages Agent Behaviours o a E e e eoe ee a Simulation Time Progression o o Using Finite State Machines for Complex Behaviours Accessing the World Model o Agent Attributes sosa es 0 0 0 ee eee ee Termination of the Agent Life Cycle Agent Logging oid ewe eee ee PED ed ee ed Performance Metric Logging o o Reproducibility lt lt recassa ammar iia E a Synchronization in Simulation o e Synchronization in PIaSMA s
42. facilities and other physical 1See http www openstreetmap org 29 30 CHAPTER 5 SCENARIO IMPLEMENTATION objects in the world These extensions are typically bundled in a file named scenario owl Details about this configuration are provided in Section 5 4 The config folder contains the scenario XML configuration s The structure of this configuration file is introduced in Chapter 4 It is possible to pool a set of scenarios which build upon the same overlapping code base into a single scenario folder In such a case for each particular scenario a dedicated XML configuration file must be placed in the config folder Note that it is not possible to configure more than one scenario in a single XML file The images folder contains images for a customized visualization of certain simulation entities such as trucks containers etc As part of the PlaSMA distribution a scenario stub is provided besides functional sample scenarios lt plasma_dir gt scenarios scenario_template When starting the development of a new scenario it is advisable to copy this folder and start developing by providing your own XML configuration the ontology based scenario specification and finally your own Java sources 5 2 Scenario Build Configuration Once you have successfully set up a new scenario workbench i e the scenario directory and started the development of your own scenario code base there is the requirement for structured compilation packagin
43. follows Windows prompt gt initdb bat lt path to postgresql_dir gt lt postgres_main_user gt 7 You will need to start the Windows command line console or an Linux Terminal first enter the directory lt plasma_dir gt and then execute the script as stated 2 1 INSTALLATION AND SETUP OF REQUIRED AND OPTIONAL SOFTWARE 7 Ubuntu prompt gt sudo i u postgres prompt gt initdb sh postgres On Windows the first argument for the script specifies the PostgreSQL installation directory Whitespaces need to be escaped with a leading The last argument specifies the main PostgreSQL user postgres by default A particular call of the script might be as follows prompt gt initdb bat C Program Files x86 PostgreSQL 9 1 postgres During the execution of the script the user is prompted to enter a password for the freshly created database user plasma In cases where the local network connection to the database is not trusted the user is required to enter the password for the PostgreSQL main user at certain stages of the setup process If the script succeeds in setting up the database PlaSMA is ready for deployment Base Configuration Method 2 pgAdmin GUI In this subsection we will guide you through PostgreSQL database setup using the graphical frontend pgAdmin III Connecting to the server After you started the Application you can see the object browser on the left hand side of the p
44. g and deployment of your code For all these steps the compilation of the Java sources the compilation of the Java class files in a single Jar archive and the installation of this archive together with the remainder of the scenario resources deployment we suggest to use the Apache Ant build system It is possible to use Ant from the command line For IDE based development Ant is also well integrated in the popular Eclipse IDE Before describing both styles of Ant usage there are some simple configuration steps required in order to set up the build system for your new scenario and your particular computer When you have started your scenario development with a copy of the scenario template mentioned above you ll see in the scenario directory the file build xml which provides generic build support The build file needs to be configured via a standard Java properties file configuration properties In order to create such a file just copy the template configuration template Afterwards you can customize the configuration to suit your system For customization only the section named user configuration needs to be edited Thereby the developer first needs to specify the location of the PlaSMA installation referred to as lt plasma_dir gt It is important to provide this location correctly such that it is possible to deploy the scenario to its correct destination If you develop more than a single scenario for deployment in PlaSMA you might fin
45. gAdmin III tool window Doubleclick an entry of the server list to establish a connection to this particular server Usually the local server shows up automatically You might have to add the server through File gt Add Server if that is not the case You will be prompted to enter the server password By default you have specified this password during the PostgreSQL database installation routine Creating anew user For database access we need a new database user with sufficient rights Therefore open the context menu right click on mouse over the list entry Login Roles of the Object Browser Choose New Login Role from the context menu A new application dialog comes up which shows the new login role to be created Figure 2 1 shows a configuration that can be used to run PlaSMA Please confirm the login role settings in this dialog to create the new login role As a result you can see this new login role in the Object Browser list view 2 New Login Role ES E New Login Role fs amp New Login Role fs Properties Definition Role privileges Role membership Variables 5QL Properties De Role privileges Role membership Variables SQL Properties Definition Role privileges Role membership Variables SQL Role name plasma o1D Password Password again eeeee Account expires Connection Limit Can login V Inherits rights from parent roles F Superuser El Can create d
46. gt optional once Configures the visualization mode optional Switches between a 2D projection planar or a 3D spheroid visualization sphere sphere is the default setting and strongly recommended lt initposition gt optional once Determines the view point of the camera when the scenario is loaded If omitted NASA World Wind will set the initial position itself based on the dimensions of the scenario infrastructure graph latitude required The latitude of the initial view point longitude required The longitude of the initial view point altitude optional The altitude of the initial view point in meters If omitted the default value of 500000 is used lt smoothiconmovement gt optional once Smoothes the movement of icons Depending on the maxDuration and the maxSpeed see Section 4 1 attributes the smoothed movement might look odd Therefore if omitted the smooth icon movement is disabled default moveSmoothly required Toggles the smooth icon movement 3 http worldwind arc nasa gov 4 Aurora does not support the 2D visualization of previous versions The sole visualization engine is NASA World Wind now However it is possible to set World Wind to a 2D projection mode of earth see below 26 CHAPTER 4 SCENARIO CONFIGURATION maxDuration required Sets the duration it will take for an icon to smoothly move to the desired position This value should be selected in regards to the maxSpeed attribute lt background gt op
47. h each execution It is also possible to use the agent s local name use the local name since the full name should change based on the host s name if no other source for varying values is conveniently accessible Random rng new Random SimSystem getRandomSeed getLocalName hashCode Using Random Generators When using a random number generator it should be restricted to a single instance of a class to avoid concurrency problems Also the order of method calls accessing the generator should be deterministic The following listing shows some examples on how to use the previously created random number generators random is assigned an integer value between 0 and 9 inclusive int random rng nextInt 10 random is assigned an integer value between 5 and 10 inclusive int random 5 rng nextint 5 percent is assigned a float value between 0 0 and 1 0 inclusive float percent rng nextFloat Using Collections If the environment is generated randomly it is imported at least for some collections to ensure that the order is always the same If for example a collection holds all trucks a logistics company manages and the company sends a message to each truck at the same time the order might be irrelevant But consider a collection containing all orders the company has to deliver The order of this collection could influence the outcome of a route planning algorithm Therefore this collection
48. he server needs to be restarted The current server state is e g displayed in the PlaSMA GUI client But usually some of the intermediate states will not be noticed because they go by so quickly These states can include e g the Scenario loaded and Initializing state 3 1 2 Parallel and Distributed Simulation PlaSMA simulations can be parallelized with multiple processors and computers To benefit from a multi processor or multi core processor machine the PlaSMA server will automatically make use of the processors available in its host system Therefore if you only use one machine there is nothing else to be done But in addition PlaSMA is also capable of distributed simulation i e simulation on multiple host computers in parallel For such cases additional settings need to be taken care of The following subsections describe these settings as well as the execution procedure to be applied Configuration PostgreSQL needs to be configured to accept incoming connections from all host machines which participate in the distributed simulation Therefore you have to allow those connections in the firewall the required port is set in the postgresql conf the default value is port 5432 and add the hosts to the pg_hba conf Assuming that all hosts are located in the subnet 192 168 100 0 the database user is named plasma and you are using the database plasma you have to add the following lines at the end of the file
49. hich is not related to the actual simulation scenario but only its implementation Both log and syslog require two parameters One has to pass the message to be logged and a corresponding LogLevel The LogLevel value can be retrieved from of the LogLevel enumeration class The given LogLevel decides whether the message is logged or is discarded In the method log there is an optional parameter for a LogCategory object As described above the log messages can be divided into different categories with different levels of granularity It is recommended to use this option since it increases the clarity of the logger output Important It is not necessary to specify all log categories in the XML configuration file If an unknown log category is used it will depend on the system wide log levels whether a specific message is logged or not One can either choose a pre defined category from LogCategory class or create a new log category on demand The latter is done by new LogCategory categoryName As a category name one has to qualify the full path of the category i e Communication Input Specifying Input as a log category name will denote a different category The agent specific logging is done automatically when the logging methods of the class SimulationAgent are used The agent specific logging suppresses the log level of any category That is if one specifies a log category whereas a separate log level is defined for the agent only the l
50. hing execution of a behaviour completely The latter is indicated by the return value of a behaviour s done method Finishing a processing cycle only prevents the behaviour from being scheduled until the next event occurs or simulation time progresses For simulation stability and causality reasons it is important to ensure that the Java source code of a behaviour does not comprise any further commands other than return subsequent to calling a finishedCycle method Otherwise unexpected side effects regarding scheduling and synchronization can occur 5 5 5 Using Finite State Machines for Complex Behaviours Advantages of a Finite State Machine JADE provides several types of behaviours for special occasions such as the OneShotBehaviour which executes its action method exactly once and terminates thereafter or the SimpleBehaviour which can be re used at any time A very powerful complex standard behaviour provided by JADE is the FSMBehaviour This behaviour is a specialization of the SequentialBehaviour Instead of executing registered child behaviours in a fixed sequence the FSMBehaviour allows to combine child behaviours to a Finite State Machine By definition a Finite State Machine is a model of behaviours that maps every possible situation of a system to a representing state Each situation might be changed by a finite number of actions accordingly transitions lead from one state to another Transitions might depend on conditions as
51. hing that is some concrete or abstract entity in a particular domain of interest Properties define relations between concepts object properties and attributes of concepts with primitive types such as strings or numbers datatype properties For instance two concepts in logistics are means of transport and cargo An attribute would be the means of transport s maximum speed or the cargo s weight A relation between means of transport and cargo is e g contains That is means of transport contain objects of type cargo Thus an object relation contains for the concept means of transport is restricted to concepts of type cargo In this relation the means of transport is called the property domain while the cargo is the property range One particular concept relation is the so called is a relation specifying a concept hierarchy of super classes and sub classes A concept C is called subsumed by a concept D if in every possible model the set of instances denoted by C is a subset of the set of instances denoted by D In contrast to conventional approaches the concept hierarchy may be predefined or logically inferred This inference is called subsumption Furthermore ontologies allow for multiple inheritance i e a concept may be the sub concept of multiple super concepts that need not form a single path in concept hierarchy The ABox defines asserted knowledge i e knowledge about instances also called individuals This incorporate
52. his method returns false or throws an ActionException the action is aborted and the executing agent is waked up By throwing an ActionException you can provide an explanation why precondition are not met 11 Java package org tzi plasma worldmodel interfaces 12 Java package org tzi plasma worldmodel actions 50 CHAPTER 5 SCENARIO IMPLEMENTATION e void checkInvariants throws ActionException Checks world model properties invariants that need to hold during whole action exection Invari ants are checked before each action exceution The method should throw an ActionException if current world model conditions do not comply with the invariants Implementation of this method is optional e WorldModelChanges execute long currentTime This method is executed by the world model to compute the effect delta of the action The effect is supposed to be computed for the time span between last action execution and the current simulation time If we reuse the vehicle example from above the method should compute the average driving speed of the vehicle and the passed distance A new ObjectPosition should be created and stored in the WorldModelChanges instance that is returned WorldModelChanges are supposed to hold all changes taking effect in the current timestamp by the given action WorldModelChanges is an read only interface Action implementations should use its implementation ObjectModelDelta e long nextExecutionTime States the simulation ti
53. ified transitions apply this register Transition MyFooBehaviour LABEL MyBarBehaviour LABEL MyFooBehaviour result SUCCESS hashCode Though hashCode should be sufficient there could theoretically be collisions If that is the case a solution would be to manually set and keep track of the integers representing each result Finishing cycles The usage of finishedCycle is nearly the same as for every other behaviour as described in Section 5 5 4 The main difference is that only the child behaviours need to call the finishedCycle method PlaSMA keeps track of the active child behaviours and a parent behaviour will finished the current cycle only if and as soon as all managed behaviours have reported that they are done for this cycle So there is no need to call finishedCycle in a FSMBehaviour itself This applies also if a FSMBehaviour contains additional FSMBehaviours as children 5 5 6 Accessing the World Model The world model provides information on the current status of the simulated physical environment with the agents and objects therein Additionally simulation agents can change this environment by 5 5 AGENT IMPLEMENTATION 49 triggering actions and events as well as by creating new simulation agents or new physical objects An agent can access the world model using method SimulationAgent getWorld This method returns the interface WorldModel which is a combination of two interfaces WorldModelQue
54. indows C Program Files PostgreSQL 9 1 data Ubuntu etc postgresql lt postgresql_version gt main postgresql conf Most of the options are already listed in the configuration file but may be disabled by the commentary symbol 4 at the beginning of the line To enable an option this symbol has to be removed In any case you should enable password encryption 10 CHAPTER 2 PLASMA SETUP password_encryption on If you wish to use more than one computer for your simulation TCP IP connections to the database must be allowed you can create a wild card for all IPs with or you allow IPs explicitly with a whitespace seperated list listen_addresses 123 456 789 1 123 456 789 3 Configuration of Access Privileges The modification of database access privileges is performed once again as the root user Per default it will only be possible to access the database locally if the login of the database user and owner of the current Linux session match The following configuration will change this state of affairs such that access is allowed both locally and from remote machines on the network using a username and password identification scheme For the sake of security both are transmitted using an MD5 Hash At the end of the file pg_hba conf change the access configuration as shown in the configuration excerpt below Database administrative login by UNIX sockets loca
55. ine Giovanni Caire Tiziana Trucco and Giovanni Rimassa JADE Program mer s Guide Telecom Italia S p A June 2007 S Bechhofer F van Harmelen J Hendler I Horrocks D L McGuinness P F Patel Schneider L A Stein and F W Olin OWL Web Ontology Language Reference Available from http www w3 org TR owl ref February 10 2004 P Davidsson Multi Agent Based Simulation Beyond Social Simulation In MABS 2000 pages 97 107 Boston MA USA 2000 Springer Verlag R Fujimoto Parallel and Distributed Simulation Systems Wiley amp Sons New York NY USA 2000 J D Gehrke Collaborative Experimentation Using Agent based Simulation In Proceed ings of the 2008 Workshop on Building Computational Intelligence and Machine Learning Virtual Organizations pages 34 38 2008 J D Gehrke and C Ober Bl baum Multiagent based Logistics Simulation with PlaSMA In Informatik 2007 Informatik trifft Logistik Beitr ge der 37 Jahrestagung der Gesellschaft f r Informatik number 109 in GI Proceedings pages 416 419 Bremen Germany September 24 27 2007 Gesellschaft f r Informatik GI J D Gehrke and A Schuldt Incorporating Knowledge about Interaction for Uniform Agent Design for Simulation and Operation In 8th International Conference on Autonomous Agents and Multiagent Systems AAMAS 2009 pages 1175 1176 Budapest Hungary 2009 International Foundation for Autonomous Agents and Multiagent Systems J D Gehrke A S
56. ing OSM data to PlaSMA Preparing a database for the PlaSMA OpenStreetMap Importer In this section the creation of a new database is described We only explain how to do this with the pgAdmin III tool Linux If you are using a Linux system you might want to create a template store the postgis schemata The template can be used again if you wish to import other data via osmosis You can create the template by executing the following commands prompt gt sudo su postgres prompt gt createdb postgis_template prompt gt psql d postgis_template f usr share postgresql 9 1 contrib postgis 1 5 postgis sql prompt gt psql d postgis_template f usr share postgresql 9 1 contrib postgis 1 5 spatial_ref_sys sql 5Until now February 2012 there exists only a stable 32 bit version of PostGIS Therefore you also have to install the 32 bit PostgreSQL version 5 4 WORLD MODEL DESIGN 39 Create the database 1 Open pgAdmin III 2 Go to edit gt create 3 Insert the name of the new database 4 Go to template and select postgis_template see Figure 5 6 5 Hit the OK button to complete the progress Figure 5 6 How to create a new PostGis database Using the PlaSMA OpenStreetMap Importer The PlaSMA OSM Importer is a tool to transform OSM data into a format that is easy to handle inside PlaSMA Scenarios At first you have to create a postgres database You can do so by simply opening 40 CHAPTER 5 SCENARIO
57. io template the situation is as shown in Figure 5 3 In order to benefit from the Eclipse Ant integration first enable the Ant view in Eclipse and arrange according to individual preferences in the active perspective Then choose Add Buildfiles from the top menu of the Ant view in order to invoke the build file selection dialogue as shown in Figure 5 3 Select the build xml file provided with the scenario template and confirm the selection You will then see a new top level item in the Ant view which represents the build file and upon unfolding reveals the distinct build targets presented in the previous section The deploy target is colored in blue as it represents the default target You may find that there are actually two flavors of Ant targets which can be told apart visually via inverted icons cf right side of Figure 5 2 These targets with a dark dot icon represent public targets which are used directly by the scenario developer The remaining targets are internal targets which can be neglected in most cases There is an option in the Ant view menu to hide these for a better overview The Ant targets are invoked via a double click Log output will be presented in the console view which is in most cases located below the main code editor window 5 3 2 Using the Ant Based Build System from the Command Line The usage of the Ant based build system from the command line both under Windows and Linux operating systems presuppose a correct ins
58. ionTimelndicator t1 t0 5 5 11 Reproducibility Usually there are some random elements in any kind of scenario For example the number of incoming orders received by a logistics company or environmental effects affecting the route a truck may take Those values could be fixed in the source code or provided through simulation configuration attributes But a main advantage of using a simulation is that the user can easily run the same scenario multiple times with different randomly generated settings to achieve a wide range of results to evaluate It is important though to ensure the reproducibility of those settings Pseudorandom Number Generation Random numbers generated by a computer are usually deterministic and the corresponding generator class therefore is called a pseudorandom number generator To achieve a deterministic behaviour the Random long constructor takes a seed argument There is a second constructor without that seed argument that relies on the current system time Don t use this for scenario random number generation The basic seed for a scenario is provided through the seed attribute in the simulation configuration see Section 4 1 This seed is accessible from every agent through SimSystem getRandomSeed and all random number generation should be based on this seed in some way The seed will change throughout multiple runs of the same scenario if more than one run is specified in the scenario configurati
59. is structured as follows In succession to this introduction Chapter 2 provides a step by step explanation of the installation of a complete PlaSMA simulation environment under Microsoft Windows and Linux The chapter also explains the configuration of both the simulation backend Section 2 2 and the client Section 2 3 Chapter 3 is then dedicated to the description how to deploy a successfully installed PlaSMA environment both for non distributed simulation settings Section 3 1 and distributed settings which involve multiple hosts that jointly host the simulation Section 3 1 2 Chapter 4 describes the way existing scenarios are set up in their respective configuration Topics which are covered here comprise amongst others the configuration of the simulation runs Section 4 1 including synchronization and domain configuration and the logging of scenarios Section 4 3 While the description of the scenario configuration so far was sufficient in order to adapt given scenarios provided for instance as standard scenarios together with the PlaSMA releases Chapter 5 is geared towards the creation of new scenarios Section 5 1 details the default structure of PlaSMA scenarios followed by an explanation of the Apache Ant based standard build and deployment process for scenarios as well as guide lines when starting work on a new scenario Section 5 4 outlines the steps to create a custom logistics world model suitable for the respective simulation t
60. itional settings for X10 import that are only stored in this files The last settings will be stored in a file named current_filter_settings iosm Other files can be named as you which but they might be easier to find if you keep the iosm file ending The following keys can be modified e OSM_FILE The full qualified path of the osm file that was used e INCLUDED_EDGES The Edge types that where used to import the data e USE_OWL Whether or not OWL should be used possible values TRUE and FALSE 8see http www vdv de oepnv datenmodell aspx for a detailed german information about the X10 data format 5 4 WORLD MODEL DESIGN 43 e OWL_FILE The full qualified path of the owl file the infrastructure should be exported to e USE_TIME_TABLE Whether or not public transport and timetable data should be used possible values TRUE and FALSE e TIME_TABLE_DIR The full qualified path of the directory where the public transport data should be saved e USE_X10 Whether or not the X10 comparison should be used possible values TRUE and FALSE e X10_DIR The full qualified path of the directory where the X10 data lies e X10_BUS_ID The ID that is used for buses in the provided X10 data e X10_TRAM_ID The ID that is used for trams in the provided X10 data e X10_ADDITIONAL_SQL_SCRIPT_FILE An additional sql file correct mappings depending on the place infrastructure you import e X10_DAY_OF_WEEK The day of the week that should be used for the import cu
61. its name enabled optional Determines whether or not the agent should be active in the scenario The default value is true If set to false the agent will be ignored 24 CHAPTER 4 SCENARIO CONFIGURATION creationDate optional Specifies the simulation time at which the agent should join the simulation This attribute uses the same format as the simStartTime attribute see Section 4 1 If omitted or set to a value before simStartTime the agent will be created at the start of the scenario run Delayed agent creation is helpful if an agent is not supposed to be active from the beginning but to be created at a specific timestamp during a simulation run Note that these agents are not created dynamically i e by other agents to react to a specific event but are created at a predefined simulation date loglevel optional Specifies an agent specific log level for each agent overriding the default scenario log level see Section 4 3 Available values are NONE ERROR WARNING INFO DEBUG and TRACE lt attribute gt optional multiple Specifies an agent property These attributes might be useful to configure simulation agents if not possible or adequate using ontology specified attributes Note that the attributes are only forwarded into the agent setup as JADE user arguments and may be interpreted therein They do not correspond to any of the simulation agent s Java class fields Attributes might be used to link a simulation agent to one o
62. kOp SFB_Logistics 08 E ld TruckTZI Logistics 02 I Y amp Truck TZI Logistics 01 E amp Truck SFB_Logistics 07 E la TruckSFB Logistics 06 E amp Truck SFB_Logistics 08 a Ce Cn cece li MeansOfTransnort 1 17 7 let Les Econs lt Le l El Copy Eo Paste Hide Unselected 1 Server State Run stopped Agents 0 Simulation Time 2009 01 0206 36 01 Syncs 1355 Run 1 1 Progress Figure 3 4 The Config Tab supports editing the agent options lt batchrun gt optional This parameter enables an alternative batch run mode The scenario parameter now referes to the parent directory of the scenario configuration e g scenario_default All configurations found in its config subdirectory will be part of the batch run Make sure to not exit the client until all requested scenario runs have finished since the client will queue any additionally scenarios configurations itself lt shutdown gt optional Specifies if the server should be shut down after all simulation runs are finished 3 3 Error Handling This section will cover errors in the programming of agents The main aspect of this section is the behaviour of the system if an exception occurs The system will automatically catch all exception and pass them to the user After that the user can analyze and resolve the error Details are following Logical errors which would not have any consequences for the simulation system will not be detected automatically If an
63. kage is called postgis The PostGis references is available under http postgis refractions net e Osmosis Osmosis is a command line Java application for processing OSM data It is especially useful for writing OSM data to databases It is also possible to filter the OSM data For example you can just import motorways motorway junctions inner city roads or other types of OSM highways Here you can download Osmosis and find an installation guide as well as a detailed docu mentation http wiki openstreetmap org wiki Osmosis In ubuntu you can install osmosis via apt the package is called osmosis Optional OSMembrane is a frontend to the Osmosis data processing tool lt helps by grouping tasks in functions and shows a visual representation of the pipeline For more information see http osmembrane de e OSM data Here you can download OSM data http www geofabrik de data download html or http downloads cloudmade com e Optional Quantum GIS Quantum GIS often abbreviated to QGIS is a Geographic Information Systems GIS application that provides data viewing editing and analysis capabilities With QGIS you can visualize the data from PostGis databases Download references and userguide http www qgis org If you are worknig under ubuntu you can simply install the packages qgis and and agis api doc from the default repositories Now you have everything you need for import
64. ke setting up individual behaviours of the agent Initial procedures like registering the agent with some service FIPA directory facilitator or an other agent should not be done in the userSetup in order to avoid concurrency problems For instance the desired service might not be running yet Below is an example of a userSetup method in which some ontological properties of the agent are read from the ontology and two behaviours are registered The agent itself is being held in an extra thread its behaviours are subprocesses 46 CHAPTER 5 SCENARIO IMPLEMENTATION protected void userSetup get represented object BasicRI myObject getReferencedObjFor OntologyObjProperty representsObject get general transport capacity will not change PhysicalObject objStatus getWorld getObjectBylD this myObject this capacity objStatus getState getDoubleValue new BasicRI OntologyService TLO_ONTOLOGY_URL storageCapacity transBeh new TruckTransportBehaviour this this addUserBehaviour transBeh procBeh new TruckProceedingsBehaviour this this addUserBehaviour procBeh 5 5 2 Reception and Processing of ACL Messages The JADE platform provides two ways for querying the message inbox of an agent There are blocking methods and non blocking methods e blockingReceive e receive Moreover derived methods can be used to setup a message filter MessageTemplate or to specify a maximum waiting tim
65. l public LoadAction BasicRI load BasicRl storageFacility long timeToLoad this storageFacility storageFacility this load load this duration timeToLoad goalPosition new RelativeObjectPosition storageFacility The init method of this Action is rather short since only the finishingTime is determined by using the getCreationTime method Override public void init this finishingTime getCreationTime duration J The next method is called after the initialization and will check the given preconditions In this case it is verified that the load is a freight object the storageFacilty is a storage facility and that both objects are in the right position by requesting those informations from the world model 13See org tzi plasma worldmodel actions LoadAction 5 5 AGENT IMPLEMENTATION 51 Override public boolean checkPreconditions WorldModelQuery wmq this getWorldModel if wmq belongsToClass load ObjectModel FREIGHT_OBJECT_CLASS_ID return false else if wmq belongsToClass storageFacility Object Model STORAGE_FACILITY_CLASS_ID return false else if wmq getPositionByID load isTransitional amp amp IcurrentPosition equals wmq getPositionByID load asTransitionalPosition return false else if wmq getObjectCoordinates load equals wmq getObjectCoordinates storageFacility return false return true The execute method is most
66. l all postgres md5 TYPE DATABASE USER CIDR ADDRESS METHOD local is for Unix domain socket connections only local all all md5 IPv4 local connections host all all 127 0 0 1 32 md5 IPv6 local connections host all all 1 128 md5 If TCP IP connections from other hosts are desired it is necessary that either the particular host or the subnetwork it resides in is granted access IPv4 connection from host 192 168 0 X host all all 192 168 0 0 24 md5 In order to finally adopt the configuration changes the PostgreSQL server must be restarted 2 2 PlaSMA Server Configuration The server configuration file at lt plasma_dir gt etc serverconfig xm1 contains global settings and is needed to run the PlaSMA server The server settings cover the message and database logging configuration The following enumeration lists all elements and whether they are required or optional and can occur once or multiple times in the configuration file An example server configuration is shown in Figure 2 4 lt server gt required once ls the top element and has one optional attribute clientport optional Sets the port at which clients can connect to the server The default port is 1099 There are two sub elements enclosed in lt server gt 81f you don t plan to run distributed simulations you should not set this option because opening your database this way may be a security risk Usually located in the
67. l may be resolved to the file lt scenario gt owl scenario owl An absolute Internet URI like http plasma informatik uni bremen de owl scenario owl may be resolved to the same file if present at lt scenario gt owl scenario owl Here is an example for the scenario ontology specification lt scenario gt owl germany_north owl lt scenario gt See Section 5 4 for more information about creating a scenario ontology 4 2 2 Simulation Agents When a simulation scenario is loaded in PlaSMA all agents and objects e g means of transport shipping containers or machine tools with their relations and properties are retrieved from the ontology cf Section 4 2 1 Some of these ontology instances might be represented by simulation agent in PlaSMA lt mappings gt required once Contains the mapping of agent descriptions i e their attributes to the corresponding Java class lt agent gt optional multiple Describes the simulation agent and maps it to its Java implemen tation This description includes various simulation related attributes as well as custom attributes accessible by the agent at runtime name required A unique name identifying this simulation agent class required Denotes the implementation of the simulation agent by its fully qualified Java class name The stated Java class has to extend the PlaSMA Java class SimulationAgent description optional Specifies a short note about the agent in addition to
68. l the PostgreSQL version which is listed in the software repository of your preferred Linux distribution In Ubuntu postgresql is the name of the required package To install the package on Ubuntu Linux type the following command on the console prompt gt sudo apt get install postgresql Make sure that postgresql client is also installed on your system Usually it will be installed with postgresql automatically You can do so by typing prompt gt etc init d postgresql status SODBC can be used to access the database with MS Excel 6 CHAPTER 2 PLASMA SETUP The output should look like this 9 1 main port 5432 online You can also choose to install the pgadmin tool if you plan to configure the database through a graphical environment instead of the shell see 2 1 1 This is done by typing prompt gt sudo apt get install pgadmin3 For full compatibility you might also want to install the following optional packages via apt e ODBC database driver package odbc postgresql e optional PostGIS see 5 4 4 driver package postgis you can do so by typing prompt gt sudo apt get install postgis odbc postgresql PostgreSQL Configuration for PlaSMA PlaSMA utilizes PostgreSQL in order to store simulation data Such simulation data can be used to analyze the model based simulation runs This in fact might help to understand in depth the computations for a p
69. lasma password plasma dbtype POSTGRESQL gt lt console enabled true colored true gt lt category name Communication loglevel NONE gt lt backtracking loglevel TRACE bufferSizeModel PER_AGENT bufferSize 100 gt lt logging gt lt server gt Figure 2 4 An Example of the server configuration lt xml version 1 0 encoding ISO 8859 1 gt lt client xmIns xsi http www w3 org 2001 XMLSchema instance xsi noNamespaceSchemaLocation clientconfig xsd gt lt PlaSMA Server Parameters gt lt server name local server address localhost port 1099 index 0 gt lt Aurora Parameters gt lt visualization enabled true gt lt client gt Figure 2 5 The default client configuration loglevel required The loglevel at which messages should be kept for backtracing bufferSizeModel required If ABSOLUTE the number of messages kept is independent of the simulation state If otherwise PER_AGENT the number of messages depends on the number of simulation agents currently in the simulation run bufferSize required The size of the buffer that holds the log messages The actual size depends on the bufferSizeModel attribute too 2 3 PlaSMA Client Configuration The mandatory client configuration located at lt plasma_dir gt etc clientconfig xml provides settings which specify the preferred PlaSMA server to which the client should establish a co
70. le for download at http plasma informatik uni bremen de download html Please unpack the contents of the downloaded archive to the directory you wish PlaSMA to be installed in In the following we will refer to this directory as lt plasma_dir gt 2 1 Installation and Setup of required and optional Software This section describes the setup of required third party software There is a short step by step guide for each of the supported operating systems In case of specific further problems please refer to the handbook for the corresponding software 2 1 1 PlaSMA Requirements The following set of third party software packages is required for the full operation of the simulation system Ubuntu is the primary Linux distribution of our core developers 4 CHAPTER 2 PLASMA SETUP e Sun Java SE 6 or newer Available from http java com download or OpenJDK 6 e optional but recommended PostgreSQL 9 1 an installation of version 8 3 8 4 or 9 0 might work as well Available from http www postgresql org download If the usage of PostGIS is planned see 5 4 4 a 32 bit version of PostgreSQL is recommended For these software packages the following installation steps apply Java Installation There are different variants of Java PlaSMA uses Oracle Java Standard Edition SE or OpenJDK other versions might work as well but are not tested by the PlaSMA Team These editions are available as software development version Java Develop
71. lt tlo edgeLength rdf datatype amp xsd float gt 120 0 lt tlo edgeLength gt lt tlo Edge gt lt tlo Edge rdf about Hamburg_Bremen gt lt tlo startingAt rdf resource Hamburg gt lt tlo directedTo rdf resource Bremen gt lt tlo edgeLength rdf datatype amp xsd float gt 120 0 lt tlo edgeLength gt lt tlo Edge gt lt physical objects gt lt tlo LandVehicle rdf about trucki gt lt rdf type rdf resource amp tlo MeansOfTransport gt lt tlo positionedAt rdf resource Hamburg gt lt tlo LandVehicle gt lt tlo LandVehicle rdf about truck2 gt lt rdf type rdf resource amp tlo MeansOfTransport gt lt tlo positionedAt rdf resource Bremen gt lt tlo LandVehicle gt lt rdf RDF gt 5 4 4 How to Import a Transport Infrastructure from OpenStreetMap Listing 5 1 OWL scenario ontology example It is quite easy to import new transport infrastructures from OpenStreetMap OSM to PlaSMA Before starting the import process you need to install the following applications and download the OSM data 38 CHAPTER 5 SCENARIO IMPLEMENTATION e PostgreSQL Version 8 4 or above including pgAdmin III with PostGis Version 1 5 PostgreSQL download and installation guide is available under http www postgresql org download Use the Application Stack Builder while installing PostgreSQL for downloading and installa tion of PostGis in Ubuntu you can install postgis via apt the pac
72. me when the action wants to be executed next if no affecting external events occur The world model will check this time in each synchronization cycle If the desired execution time is reached the world model will execute the action Please note that the actual execution time may be delayed due to time progression granularity as configured by minSyncDist scenario property cf Sect 4 1 1 e WorldModelChanges transitionState This optional method is called if an incomplete action is not executed but was registered or executed in the previous world model update It gives the action the possibility to indicate a transitional world model state of the action e g to move objects into Transitional0bjectPositions e boolean isCompleted This method indicates whether the action has been completed yet This property has to be set by the action programmer and is evaluated by the PlaSMA world model Complete actions will not be scheduled again An Action Implementation Example The Load Action After the description of the important methods provided by the Action interface this paragraph will give a small example on how to implement those methods by looking at the LoadAction This Action is supposed to store a package into a storage facility thus it requires the information which load to store into which storageFacility lt also takes into account how long the loading process will take timeToLoad public class LoadAction extends Action a
73. med test owl and imports the TLO using the lt owl imports gt tag The DOCTYPE section lines 2 11 just defines standard abbreviations as XML entity definitions that for instance can be used as prefix for identifiers in other XML namespaces when used in XML attribute values or tag contents lt rdf RDF gt is the actual root XML tag Its attributes in lines 14 22 define abbreviations too But these may be used for identifiers of XML tags and attributes only not their content or values The following lines 29 60 denote the actual scenario OWL description with instance and property assertions The given scenario consists of two locations Bremen and Hamburg a bidirectional traffic link between them and two trucks 3 The scenario ontology s default file name is scenario owl You may also use other file names depending on the XML scenario setting 4 Available for free from http protege stanford edu download download html 5 4 WORLD MODEL DESIGN 37 lt xml version 1 0 gt lt DOCTYPE rdf RDF 1 gt lt ENTITY owl http www w3 org 2002 07 0w1 gt lt ENTITY ow111 http www w3 org 2006 12 owl11 gt lt ENTITY xsd http www w3 org 2001 XMLSchema gt lt ENTITY owliixml http www w3 org 2006 12 owl11 xml gt lt ENTITY rdfs http www w3 org 2000 01 rdf schema t gt lt ENTITY rdf http www w3 org 1999 02 22 rdf syntax ns gt lt ENTITY tlo http plasma informatik uni brem
74. ment Kit JDK and as plain run time version Java Runtime Environment JRE It is pre installed on several systems But you have to check that it is at least version 6 If you want to implement new agents the JDK is needed If the version of the installed JRE or JDK is below Java 6 an update is necessary Java installation on Windows Hosts If there is no appropriate Java installed download the Java installer JDK or JRE for your Windows platform from the Website stated above Just execute the Java installation program and follow the instructions There are no specific installation options required by PlaSMA In any case you have to make sure that Java is contained in the Windows program search path listing environment variable PATH The listing separated by semicolons must contain an entry like C Program Files Java jdk1 7 0_10 jre bin or C Program Files Java jre7 bin This path must correspond to the location where your Java JRE or JDK is installed Add the correct path if that is not the case Java installation on Ubuntu Hosts If you are using Ubuntu you can simply install the newest version of OpenJDK JDK or JRE from the Ubuntu repositories prompt gt sudo apt get install openjdk 7 jre or if you need the JDK prompt gt sudo apt get install openjdk 7 jdk PostgreSQL Installation PostgreSQL is the database used by PlaSMA to store your simulation data If you do not intend to evaluate
75. moving the whole XML tag The default value is true lt filterobject gt optional multiple States that an ontology ID should be explicitly taken account of Usually only the most specific and some predefined ontology concepts are considered by the world model while all other concepts are filtered out This might be useful if a specific concept should be displayed by the visual client 4 2 4 Online Scenario Visualization The PlaSMA GUI Aurora can handle scenario specific visualization settings These settings include e g special images for certain objects or object types as well as 3D scenery properties such as an initial camera position The PlaSMA GUI uses the NASA World Wind framework for its 3D planetary visualization lt visualization gt optional once The embracing XML tag for all visualization settings enabled optional Allows to deactivate visualization temporarily without the need to remove the XML settings lt images gt optional once Denotes an XML sequence of generic image mappings lt image gt optional multiple A single image mapping key required The unique image identifier The key may e g correspond to the ontology ID local or with namespace of any object or the ontology ID with namespace of an object type src required The URI of the image file A relative URI will be resolved relative to the scenario directory or if not found there relative to PlaSMA resource directory etc lt worldwind
76. name gt Ubuntu prompt gt createuser U postgres s P E lt user_name gt Then we will create a new database dedicated to PlaSMA Enter and execute the following command This command takes the aforementioned user name as an argument You may choose an arbitrary database name by default it is plasma Windows prompt gt createdb exe U lt user_name gt lt database_name gt Ubuntu prompt gt createdb U lt user_name gt lt database_name gt At last the database structure is being generated To build up the database structure we use a create script Enter and execute the following command This command takes the full path to our create script as an argument as well as the name of PlaSMA database Windows prompt gt psql exe U lt user_name gt d lt database_name gt f lt plasma_dir gt sql plasma_datamodel sql Ubuntu prompt gt psql U lt user_name gt d lt database_name gt f lt plasma_dir gt sql plasma_datamodel sql Additional Configuration for Distributed Simulation To record data during a distributed simulation session it is necessary to modify the configuration of the PostgreSQL database system After a successful installation you need to edit the file postgresql conf as user with administrator root privileges The location of this file depends on your operating system If you use the default installation Directory this might be W
77. nced0bjsFor Ontology0bjProperty BasicRI company getReferencedObjFor OntologyObjProperty operatesFor The GUI scenario configuration editor is capable of suggesting attributes created this way but needs to associate them with an agent Therefore the enum implementing the ConfigKeyInterface should be linked to all agent classes that support that attributes through the AgentConfiguration annotation MSee Section 4 2 2 for details Borg tzi plasma toolkit forwardingagency agent Agent_ForwardingAgency lorg tzi plasma control ConfigKeyInterface Vorg tzi plasma control SimulationAgent Ontology0bjProperty ort ane ono 10 11 12 13 14 15 16 17 18 5 5 AGENT IMPLEMENTATION 53 AgentConfiguration CustomAttribute class ForwardingAgencyAttribute class public class Agent_ForwardingAgency extends SimulationAgent L Implementing the ConfigKeylnterface Though it is possible to implement the ConfigKeyInterface in any class it is recommended to use an enum The following listing contains an excerpt of the CustomAttribute implementation package org tzi plasma toolkit forwardingagency agent import org tzi plasma control ConfigKeyInterface import org tzi plasma control ConfigAttribute import org tzi util BasicRl public enum CustomAttribute implements ConfigKeylnterface run_mode ConfigAttribute of Type String class withValues DECENTRAL_PLANNING Mandatory in configuration depo
78. nd the configuration needs to be redistributed manually to the other hosts 18 CHAPTER 3 RUNNING PLASMA Depending on the size of the scenario number of agents size of the traffic infrastructure the loading of the scenario may take a few seconds or even longer After the scenario has been loaded the simulation can be started paused and stopped with the Play Pause and Stop button in the main toolbar Once the scenario is loaded you can look at it from 3 different perspectives The Server Log see Figure 3 1 shows recent messages from agents and the simulation system itself In this view you may use the menus Log Levels Agents and Log Categories to filter which types of messages you want to be displayed The filter text field is another convenient way to look for specific messages If Clear on next run is selected the Log will be cleaned up at the beginning of each run for the sake of clarity The Sequence button can be used to toggle if the messages should be shown in ascending or descending order A PlaSMA Aurora ej SE Server Options Ej Select Scenario D amp Server Log WorldWind Visualisation 4 Scenario Configuration Clear on next run Sequence Start with oldest Cl Log Levels Q Agents E Log Categories Enter Filter Text here SE Server State Scenario ready Agents 0 Simulation Time 2009 01 01 08 00 00 Syncs 0 Run 1 1 Progress 0 hm Figure 3 1 The Log T
79. nd usePlanar which result in a 3D or 2D representation respectively If omitted the default value use3DGlobe will be used mode required Sets the display mode Available parameters are use3DGlobe and usePlanar lt max_zoom_on_startup gt optional once Specifies the maximum camera elevation when starting a scenario to avoid the camera moving too close to the ground especially in scenarios with only one node The default value is 50000 lt localization gt optional once Changes the localization of the client language required Sets the language specified by a language code followed by a hyphen and a country code For example en US or de DE The default language if this parameter is omitted is English en US 12Conform to RFC 3066 Running PlaSMA This chapter is concerned with the general usage of the PlaSMA multiagent based simulation environment in both local and physically distributed settings Special aspects like scenario implementation and evaluation are addressed in subsequent chapters The general usage of PlaSMA in this chapter covers the execution of the PlaSMA server potential remote controllers as well as PlaSMA s command line or GUI clients All start scripts are located directly in lt plasma_dir gt 3 1 PlaSMA Simulation Server The start of a PlaSMA server depending on the host operation system is done with the following command Linux server master sh lt password gt jade lt jade args gt
80. nnection Figure 2 5 shows an example for a client configuration Editing the connection informations is fully supported by the visual client itself see Section 3 2 1 Choose Server Options gt Connect to Edit Connections from the main menu panel The following enumeration lists all available configuration settings with a short explanation lt client gt is the top element without any PlaSMA specific attributes The client element may contain different sub elements as follows lt server gt optional multiple Each entry defines a server available in the client s Connect to menu If no server is specified the client automatically generates a default server entry with the address localhost and port 1099 name required The displayed name of the server address required Declares the DNS name or the IP address of the PlaSMA server port required Specified the port the PlaSMA server is listening on 2 3 PLASMA CLIENT CONFIGURATION 13 index required An index used for internal ordering of all available server entries The lower the number the higher up the entry will appear in the menu lt visualization gt optional once Toggles the visualization of the scenario The visualization will be enabled by default if this setting is omitted enabled required lf set to false the visualization of scenarios will be disabled lt displaymode gt optional once Chooses the display mode Currently two settings are supported use3DGlobe a
81. noting two concepts e g airplane and ISO container as disjoint The Logistics Ontologies The simulation system provides five OWL ontologies as TBoxes for world modeling which are briefly described in the following list e TLO The top level domain ontology for logistic scenarios TLO specifies general types of physical objects the basics of traffic infrastructure organizational membership ownership of physical objects and software agents providing abstract services All other ontologies import this ontology for extension e Trans The ontology for transport specific issues For instance this ontology defines several types of nodes and links in the traffic infrastructure as well as a number of different types of vehicles and transport containers e Prod Production logistics is handled by this ontology It specifies machines and production orders with their respective properties e Com This communication technology ontology defines communication and computation devices with properties such as radio coverage power supply type and computational power e g for mobile agents e Goods The goods ontology provides a general schema to classify goods For instance classification may differentiate physical or non physical good e g property rights packaged good or bulk good food or non food dangerous good or not phase of matter etc The above ontologies are neither considered complete nor mandatory That means the system use
82. nstance public static enum result SUCCESS FAILURE ANY_OTHER_RESULT i Basically the jade core behaviours FSMBehaviour provides all necessary methods to create a Finite State Machine As a Finite State Machine has a first state and at least one last state registerFirstState and registerLastState allow to define start and stop behaviours Any other states in between can be registered with the method registerState Required argu ments are the behaviour that will be represented by the state and a unique String label for this state To keep the code of the FSMBehaviour independent from the managed behaviours it may be useful to keep the label for each state as a static variable within the respective behaviour class this registerState new MyFooBehaviour this myAgent this getDataStore MyFooBehaviour LABEL this registerState new MyBarBehaviour this myAgent this getDataStore MyBarBehaviour LABEL To implement dependencies between the states transitions have to be registered either as default transitions e g if the following state will always be the same or as a conditional transition Every state may have various outgoing transitions For consistency every transition needs a unique condition as discussed before It is possible to register a default transition and conditional transitions from the same state In such a case the default transition will be used when none of the other explicitly spec
83. of nodes are e g trans TrafficJunction like a motorway interchange tlo Source and tlo Sink to denote origin and destination of goods as well as trans Seaport and trans Airport where edges of different types come together Figure 5 5 gives an overview on the most important edge types with the corresponding concept hierarchy Refer to the ontology trans owl for a comprehensive description of all predefined types The following extract of a sample scenario ontology shows the corresponding XML syntax of a graph specification lt tlo Vertex rdf about Bremen gt lt tlo latitude gt 53 0759 lt tlo latitude gt lt tlo longitude gt 8 8073 lt tlo longitude gt lt tlo Vertex gt lt tlo Vertex rdf about 4Hamburg gt lt tlo latitude gt 53 5506 lt tlo latitude gt lt tlo longitude gt 9 9933 lt tlo longitude gt lt tlo Vertex gt lt tlo LandRoute rdf ID Hamburg_Bremen gt lt tlo startingAt rdf resource Hamburg gt lt tlo directedTo rdf resource 4Bremen gt lt tlo edgeLength gt 120 0 lt tlo edgeLength gt lt tlo LandRoute gt lt tlo LandRoute rdf ID Bremen_Hamburg gt lt tlo startingAt rdf resource Bremen gt lt tlo directedTo rdf resource 4Hamburg gt lt tlo edgeLength gt 120 0 lt tlo edgeLength gt lt tlo LandRoute gt 36 CHAPTER 5 SCENARIO IMPLEMENTATION Static Objects and Graph Nodes The PIaSMA ontology considers physical objects disjoint from traffic infrastructure a
84. ogging category for which a loglevel is to be set loglevel optional The special loglevel for this category Inherits from parent category by default lt backtracking gt optional once Configures the tracing of errors during simulation runs The system can be configured to temporarily store a certain amount of protocol messages that would normally be discarded regarding their log level In case of an error they will be logged and may be helpful in tracing back the error The attribute loglevel limits which protocol messages are temporarily kept The number of messages has to be configured by the bufferSizeModel and bufferSize settings If the model is ABSOLUTE a maximum of bufferSize messages is temporarly stored whereas the PER_AGENT model is allowed to store bufferSize messages per agent Due to the distributed architecture of the simulation these settings apply for every connected computer and not for the simulation cluster as a whole That is because the traceback information are managed locally on every participating computer 11See section 4 3 for more details on log levels 12 CHAPTER 2 PLASMA SETUP lt xml version 1 0 encoding ISO 8859 1 gt lt server xmins xsi http www w3 org 2001 XMLSchema instance xsi noNamespaceSchemaLocation serverconfig xsd clientport 1099 gt lt scenariodir path scenarios gt lt logging loglevel INFO gt lt database enabled true host localhost name plasma user p
85. on agent parameter specifies the agent the performance metric will be logged for For example there might be multiple trucks logging their driven kilometers using the same performance indicator key that can only be distinguished by their corresponding agent A PILogger created this way can be further split up into different contexts For example the toolkit s default scenario uses one agent to manage all storage facilities In this case the key and owner are the same for every facility To be able to distinguish them the withContext String method can be chained to the previously created logger PlLogger pilogger PlLogger forOwner agent To finally log a value the PILogger takeMeasurement PIKeysInterface T method must be called on the previously created PILogger instance providing the corresponding performance indicator key and the value to be logged The values are logged once per cycle so assigning multiple values within one cycle only gets the latest value logged The generic class parameter T is set to the class of the first value logged for a performance indicator key So if for some reason an indicator should be able to log different classes the first value must be cast to a class that all values share and implements the Serializable interface long t0 System currentTimeMillis path algorithm getShortestPath start destiantion long t1 System currentTimeMillis pilogger takeMeasurement computat
86. on s repeats attribute The seed of a run beyond the first is always based on the previous run s seed Therefore it is possible to repeat a specific range of runs from a complex series If all runs should use the same seed again a custom attribute could be used or all runs have to be treated as single runs and started with only one repeat each 19Spaces and characters are not allowed 20See java util Random 56 CHAPTER 5 SCENARIO IMPLEMENTATION Creating Own Seeds Having all Random long generators use the same seed is often inconvenient So it is necessary to generate new seeds based on the scenario s base seed One very important aspect to keep in mind is that every factor influencing the random generation needs to be deterministic If the same scenario run concludes with different sync counts it is most likely that somewhere non deterministic numbers are generated It might be handy to provide a fixed seed generation value for an agent generating its own numbers and combine that value with the base seed Random rng new Random SimSystem getRandomSeed x agent_seed_attribute Or if multiple instances of the same agent are used provide them with some kind of id and use that for calculating different seeds per instance Random rng new Random SimSystem getRandomSeed x id If using Java s hashCode method make sure it is adjusted to the needs of random number generation and doesn t change wit
87. onfiguration is explained in Chapter 4 PlaSMA scenarios coarsely exhibit the content structure as shown in Figure 5 1 Each scenario is self contained such that it is possible to build only a particular scenario of interest in place as all necessary resources are included to deploy i e install the scenario for use with the PlaSMA simulation system Note that using identical fully qualified class names in different scenarios might cause problems since only the first occurrence of that class will be loaded This applies to external libraries used in scenarios as well All scenario source code is placed under a certain source folder src by default i e the build process should not depend upon sources external to the scenario folder All custom libraries which are required for the compilation and deployment of a particular scenario are to be placed in the library folder lib by default The owl folder contains the ontology based specification of the simulation world model This model may contain scenario specific schema extensions to the general OWL ontologies tlo trans comm prod which are provided with the PlaSMA system These extensions comprise the introduction of new concepts and relations which are not covered by the general ontologies e g a groceries ontology The other and most important part of the ontology based world model is the concrete instance level configuration of the scenario such as the transport network production
88. our file system for the desired file or you type in the desired URL After you provided the information above you can start the import by clicking on start button on the left site of the main window The process might take a while You can see information about success or failures in Log panel Filtering relevant data The Filter panel provides settings for filtering imported data These settings can be used to reduce the complexity of the osm data without having effects to granularity of the graph You can start the progress by clicking on the start button on the bottom of this panel The start button triggers a mechanism that removes all nodes that have only one incoming and outgoing edge There are some options to change the default behavior of the mechanism e Export to OWL You can export all data to an ow ontology file To do so check the box under Export infrastructure to Ontology If you decide to export the ontology to owl you have to specify a path where the infrastructure should be saved in the field next to the checkbox e Include public transport infrastructure and timetables If you wish to use information about public transport you can store this data as a csv file For this you have to specify a target You can download osm data from various sites For example http www geofabrik de data download html or http downloads cloudmade com 42 CHAPTER 5 SCENARIO IMPLEMENTATION directory in the field time table directory
89. peated Scenario Runs 1 Start Time 01 01 2009 08 00 00 Max Speed E Unlimited 10000 0 x Real Time Seed Value 1230796800000 End Time 11 01 2009 08 00 00 Synchronization Options Min Sync Dist 00 00 20 F Max Sync Dist 00 15 00 E Timeout 20 Seconds Optional Settings to Load the Transport Infrastructure from Database Database URL Login Name Login Password 1 Server State Run stopped Agents 0 Simulation Time 2009 01 0206 36 01 Syncs 1355 Run 1 1 Progress SEOs Figure 3 3 The Config Tab supports editing the scenario options Linux prompt gt client cli sh server lt server gt port lt port gt lt scenario gt prompt gt client cli bat server lt server gt port lt port gt lt scenario gt The startup parameters are as follows Windevserver gt optional The address of the server the client should connect to For local simulation this parameter can be skipped since localhost is the default value lt port gt optional The port the server listens on The default value is 1099 lt scenario gt The name of the scenario to load This is the attribute name from the scenario configura tion If this name contains any whitespace put it in quotes Using as name will list all available scenarios 20 CHAPTER 3 RUNNING PLASMA GB PlaSMA Aurora a elo EE Server Options Ej Select Scen
90. peed might differ depending on machine power and scenario specifics seed optional The random seed for the scenario This value can be used to initialize random number generators PRNGs for the scenario to achieve a deterministic behavior of the simulation The value is given as long integer If this attribute is missing a fixed seed is used The seed can be read out by the agent with SimSystem getRandomSeed The usage of this random seed is very important to the reproducibility of scenarios as well as to the reproducibility of results see Section 5 5 11 lt description gt optional once A description of the scenario to be displayed in the client 4 1 1 Synchronisation Configuration lt sync gt required once This subelement defines the time control and synchronization of the simulation see Section 5 6 for further explanations It contains the following attributes minSyncDist required Specifies the minimal progress of simulation time in a synchronization step This allows for grouping multiple agent s execution in the same synchronization cycle Otherwise it is most likely that each agent requests its own synchronization step with all other agents being idle E g having two agents requesting a synchronization in 5 and 10 seconds respectively With a minSyncDist value of 1 second two cycles are completed with only one agent being active If set to 20 seconds only one cycle with both agents running their tasks is executed A red
91. pped The input format of maxSimLength consists of a number and a time unit The time unit can be one of the following e ms milliseconds e s seconds sec seconds e min minutes e h hours e d days 21 22 CHAPTER 4 SCENARIO CONFIGURATION e y years If no unit is specified the number is interpreted as milliseconds Another time unit which is not listed above leads to an load error when loading the scenario simStartTime optional Specifies the simulation time at which the simulation scenario starts This attribute can be specified either as UNIX time stamp thus as milliseconds since January 1st 1970 or as a date in the YYYY MM DD HH mm ss format If this attribute is missing the current system time is used as starting time for the scenario maxSpeed optional This attribute should be set if the scenario is visualized too This attribute specifies the maximal rate of simulation time to computation time A value of 1000 means that the simulation time elapses not faster than 1000 times real time A value of 1 would mean that it will pass exactly as fast as the real time accordingly Values less than 1 but greater than 0 are allowed too which would corresponds to slow motion Values less or equals to 0 will deactivate this attribute and the simulation time elapses as fast as possible The default value is 1 The speed limit is realized by a delay of the next synchronisation step The actual simulation s
92. r is free to use or not use one or more of these ontologies for a scenario and may also adapt or extend them for individual needs There is only one exceptions to this e The traffic infrastructure and physical world model is considered to be a directed graph of nodes and edges with certain mandatory properties such as length and mandatory types cf Sect 5 4 1 The easiest way to keep the system running correctly is not to modify the TLO Ontology Entities Each named resource in an OWL ontology concept property or instance has a unique resource identifier In order to guarantee cross ontology uniqueness resource identifiers have a namespace as prefix which is the URI of their ontology extended by a character Ontologies may define shortcuts for namespaces of imported ontologies By default the ontology s short name in lower case e g tlo is used in PlaSMA To indicate that a resource belongs to the top level ontology we write e g tlo PhysicalObject or tlo SoftwareAgent The scenario ontology namspace has the default abbreviation scen These abbreviations are provided as String constants in Java class OntologyService In PlaSMA the URI of a resource is represented by the Java class BasicRI i e basic resource identifier This class only encapsulates two character strings namespace and local ID and is optimized for handling in PlaSMA 5 4 WORLD MODEL DESIGN 35 tlo StructureElement Figure 5 5 Hierarchy of the mos
93. r more specific ontology instances e g by using some kind of operatesFor or representsObject property with an ontology ID as value key required The unique key of the attribute value required The value of the attribute in string representation description optional A short description of this attribute lt attributeset gt optional multiple Extends lt attribute gt to support multiple values per key key required The unique key of the attribute description optional A short description of this attribute lt entry gt required multiple A list of one or more values associated with this key value required One value of the attribute in string representation The following listing shows an example agent mapping configuration lt mappings gt lt agent class org tzi plasma toolkit forwardingagency agent Agent_ForwardingAgency description T ZI Logistics Forwarding Agency enabled true name Management T Zl_Logistics gt lt attributeset key available_locations gt lt entry value http plasma informatik uni bremen de owl shared_infrastruture owl Storage ACME_Air_Assembly gt lt entry value http plasma informatik uni bremen de owl shared_infrastruture owl Storage Bremen gt lt attributeset gt lt agent gt lt agent class org tzi plasma toolkit truck agent Agent_TruckManagement creationDate 1230800400000 description TZI Truck Management Agent 3 enabled true name TruckOp TZI_Logistics 03 gt
94. red simultaneous although they would not be in physical time GSW08 By contrast in discrete event simulation DES time steps bridge the gap until the next event occurs Thus DES is the dominant approach in most cases Thus PlaSMA uses a time model with discrete steps resulting from scheduled agent actions or explicit agent wake up requests 5 6 1 Synchronization in Simulation Agents in MABS can be simulated on distributed platforms which may differ regarding their computational power Even on one platform the local virtual time of agents may diverge if their computational demands differ For instance consider an agent passing a message to another agent that is advanced in its local virtual time The recipient of such a so called straggler message might have taken other decisions if it were aware of that message on time This is denoted as the causality problem Fuj00 In order to guarantee correct and reproducible simulations the simulation system has to ensure that agents process events in accordance to their time stamp order This problem is addressed by synchronization which can be either optimistic or conservative In general progression of local virtual time is not restricted for agents in optimistic synchronization This allows 58 CHAPTER 5 SCENARIO IMPLEMENTATION executing simulations efficiently since fast processes do not have to wait for slower ones Whenever an agent receives a straggler message with time stamp t a rollba
95. rom the returned array final Object args getArguments final Map lt String List lt String gt gt params Map lt String List lt String gt gt args 0 5 5 8 Termination of the Agent Life Cycle The recommended way to control termination of a simulation agent is to override the method isAgentDone in the class extending SimulationAgent This method always returns false in the default implementation The method is evaluated at the end of each agent processing cycle The agent is properly terminated and deregistered from PlaSMA if it returns true For termination doDelete is executed Worg tzi plasma toolkit forwardingagency agent CustomAttribute 54 CHAPTER 5 SCENARIO IMPLEMENTATION If isAgentDone O is insufficient to handle situations when the agent shall be terminated you can call this method directly in your behavior code as well When doing so you must not call finishedCycle O for this behaviour anymore Otherwise the agent might deregister with PlaSMa twice which will cause an error Either way you need to take care though that an agent that registered with the directory facilitator agent DF needs to deregister before termination In order to do such clean up work override the method onTermination of SimulationAgent 5 5 9 Agent Logging One can use log method or syslog method of SimulationAgent class to log messages Where syslog should be used if a system message is to be logged w
96. rrently only one day is cloned for every day of week NODE_ID The 9 digit node ID of a node that is in the main graph all edges that can not be reached from this node will be removed Optional Visualization with QGIS You can visualize your PostGis data with QGIS to do so follow the steps below e Open QGIS e Create a new Layer see Figure 5 9 Z Quantum GIS 1 7 1 Datei Bearbeiten Ansicht Layer Einstellungen Erweiterungen Raster Vektor Hilfe EELEE BRC C RRP ES carros 7 RAARKXAAA LO e Oe Ot A A O Bu Uf la A gt ONDGBLHOOC VARIS ORR T layer Ex Figure 5 9 Create a new Layer e Connect QGIS to PostGis see Figure 5 10 e Select the data that you want to visualize see Figure 5 11 e It is important that you change the datatype of the ID columns from bigint to integer in your databases We recommend to do this with pgAdmin III that is installed with PostgreSQL Manipulate the scenario configuration At last you have to specify in your scenario configuration xml file to use an ontology file or the database for representing the transport infrastructure In the later case you also have to configure the connection properties in the following way 44 CHAPTER 5 SCENARIO IMPLEMENTATION Create a New PostGIS connection 2 Y Y Y Figure 5 10 Connect to PostGis Add PostGIS Table s E public all_stops_by line POINT coordinates AS public junction_node POINT geom AS POINT public result_jun
97. ry for obtaining information and WorldModelAdmin to alter the model Among others WorldModelQuery includes the following methods e PhysicalObject getObjectByID BasicRI objectID Returns a descriptor of a physical object in the world model identified by its object ID The descriptor contains object properties but is not adequate to actually change the object properties e ObjectPosition getPositionByID BasicRI objectId Returns the given object s current position e DirectedGraph getGraph Returns a directed graph as the scenario s underlying traffic infrastructure BasicRI is a simple datatype representing an unique resource identifier URI for a resource object in the simulation model It consists of two strings for the namespace and the local ID respectively Most of the methods in this interface are self explanatory For a complete and more detailed description of the interface please refer to the Javadoc API documentation of PlaSMA World Modifications with Actions To alter objects in the world model i e by moving objects so called Actions have to be defined for the scenario A couple of base and example actions are already included in PlaSMA To trigger an action the simulation agent has to register it with the world model by calling WorldModel registerAction action agentID Until their completion actions will be re evaluated by the world model For every progression of simulation time all actions will be evaluated
98. s This will result in a suitable distribution of the workload on all computers and therefore a maximal performance of the whole simulation The agents stay always on the computer they were started on In case of intensive agent communication and very different agents this may lead to a suboptimal workload distribution 3 2 PlaSMA Clients 3 2 1 Visual Client The control visualisation and visual feedback respectively is done by the PlaSMA client GUI Aurora It is started with the following command Linux aurora sh Windows aurora bat To connect the client with the server choose Server Options Connect to in the window menu From startup only the local server is available You may create new entries for any server you wish to connect to by selecting Server Options Connect to Edit Connections When stating a host name for the PlaSMA server instead of an IP it is necessary that the client can resolve that host name e g by adding it to the etc hosts Linux or C Windows System32 drivers etc hosts Windows file After a successful login to the simulation server you can select a scenario from the dropdown menu Select Scenario The menu contains all valid scenario configurations in your scenario folders usually lt plasma_dir gt scenarios unless specified otherwise see Section 2 2 21f the scenario configuration got changed through Aurora see Section 3 2 1 that changes are only present on the master server a
99. s assertions like some particular object being an instance of a concept concept assertion or having some property value role or property assertion While a property in the TBox defines domains and ranges the ABox defines property values For instance an ABox may contain the assertions that truck1 is an instance of concept means of transport and in relation contains an instance cargol Again ontologies allow for logical inferences here We did not state that cargol is an instance of concept cargo 34 CHAPTER 5 SCENARIO IMPLEMENTATION but we may infer this because the range of property contains for concept means of transport is cargo The inference of determining if an individual is an instance of some concept is called instance checking The inference of determining all of a concept s instances is called retrieval Realization is the inference of finding the most specific concept s of some particular instance The above inference tasks for TBoxes and ABoxes are not complete In particular one important inference in TBoxes is satisfiability of concepts i e checking whether a set denoted by a concept may be non empty in some model This helps you detect if your designed world model is logically consistent For instance the check will fail if you model a type of airplane that is also asserted or inferred to be an ISO shipping container because no such things are allowed to exist in the ontology Such mutual type exclusions are expressed by de
100. s given by elements in the directed graph Thus physical objects are located at structure elements but are no structure elements themselves That is in general a structure element should have no function besides representing locations and connections between them and a static object like a storage facility is located at some node and represents the ability store items there Exceptions from that rule are possible and up to the designer of the scenario 5 4 2 World Model and Agents All autonomously acting entities in a simulation scenario are represented by their corresponding Java class so it is not necessary to model simulation agents at all However it is possible to model those agents if that is required for some reason Although it may seem reasonable to equate physical objects e g a truck or services with simulation agents it is a matter of confusion Agents are e g humans organisations or artificial entities An agent may represent objects and decide in the interest of the object s owner but in general it is no physical object itself The exception would be humans or robotic agents which can be viewed as physical objects as well as agents Furthermore one has to distinguish an agent s attributes and the attributes of the object it acts for or the service it provides In particular an agent s position if any is not necessarily equal to the position of the physical object it represents Also it may represent more than one objec
101. s s s sp oo o 2 CONTENTS Introduction The software and agent platform PlaSMA originally developed in the context of the Collaborative Research Center 637 SFB 637 at the University of Bremen provides means for the simulation of large scale logistic scenarios Agents may represent autonomously acting entities such as means of transport and cargo Depending on their respective specification these agents constitute either the intelligence of a single entity in the simulation environment or they manage groups of entities in the simulation environment A further class of agents is that of service agents which have secondary logistic functions and thus provide for instance traffic information or contract brokering services Besides the simulation backend and a graphical visualization client the PlaSMA installation package comprises a selection of example logistics scenarios that have been developed for demonstration purposes The document at hand provides detailed information about the process of setting up the PlaSMA simulation environment and the compilation configuration of custom application scenarios Besides the simulation model which rests at the core of the PlaSMA simulation backend is introduced i e the synchronization model and the server world model In the end the document describes the steps that developers must take in order to successfully implement new agents based on the PlaSMA API The remainder of this user guide
102. scenario s default log level Available values are NONE ERROR WARN ING INFO DEBUG and TRACE lt logcategory gt optional multiple Specifies a log category name required The name of the log category loglevel required The log level of this category Available values are NONE ERROR WARNING INFO DEBUG and TRACE The following listing shows an example logging configuration excerpt lt logging loglevel INFO gt lt logcategory name Agent loglevel NONE gt lt logcategory name Agent Planning loglevel DEBUG gt lt logging gt Scenario Implementation 5 1 Overview of the General Scenario Structure A PlaSMA scenario consists of three basic parts e The world model structure is specified by OWL ontology files Therein one can define which kinds of objects and agents are part of the scenario as well as the network infrastructure they are situated in It is also possible to import the infrastructure from OpenStreetMap see Section 5 4 4 for details e The simulation agents are the actual simulation processes that implement the behaviour of physical objects and or software agents Simulation agents are implemented as Java classes e Simulation specific scenario settings are made in an additional XML file Essentially in this file the user can map objects to simulation processes i e simulation agents There are additional settings for simulation time management visualization and logging The settings c
103. se class method addBehaviour if needed When using unsynchronized behaviours please ensure that the agents do not communicate with each other off time their synchronized cycles This would distort the simulation results or at worst lead to unplanned downtime of the simulation It is recommended to instantiate initial behaviours within the userSetup method Behaviours are being processed sequentially in the order of their registration time There are two ways to remove a synchronized behavior When a behaviour is finished done true it is automatically removed from the scheduler If it should be removed explicitly please call the method removeUserBehaviour 5 5 4 Simulation Time Progression In order to realise time progression in simulation agents indicate that they finished their processing cycle for the current simulation time This is accomplished by requesting a new time event from the simulation controller This request is sent automatically by each agent when all active behaviours of that agent finished the processing cycle In order to indicate that a behaviour has finished computation for one cycle it calls the myAgent finishedCycle UserBehaviour b long nextSimTime method The first parameter refers to the behaviour itself and the second parameter provides the simulation time at which the behaviour has to be scheduled again as long as no other event occurs before Note that finishing a processing cycle differs from finis
104. should be sorted before applying the route planning algorithm To be able to sort a collection it is required that it can be sorted i e it should be a list or be able to sort itself This might require to convert the given collection into a suitable format The values of the collection also need to be comparable i e they have to implement the java lang Comparable interface Alternatively or in case a different order instead of the default ascending order is desired a java util Comparator can be passed as optional second parameter The collection then could be sorted using the java util Collections sort method or is already sorted by itself in case of a SortedSet 21See java util List 22See java util SortedSet 5 6 TIME MANAGEMENT AND SYNCHRONIZATION 57 public void planning Collection lt Order gt unsortedOrders List lt Order gt orders new LinkedList lt Order gt unsortedOrders Collections sort orders new Comparator lt Order gt Override public int compare Order a Order b return a value b value Equals and Hash Code Since agent communication might involve the serialization of objects it is advisable to adapt the equals method The default implementation for Object only returns true if both objects refer to the same object which is not the case after serialization There might also be other cases where this could be an issue e g reading a string based configura
105. sions of PostgreSQL are available as a Windows Installer package from the PostgreSQL website at the downloads section There are two kinds of installer packages available for Windows operating systems Generally it is recommended to choose the one click installer package The one click installer package is designed to be as straightforward as possible and the fastest way to get up and running with PostgreSQL on Windows The following installation instructions regard to the one click installer package In case you are about to run PostgreSQL on a Windows operating system which is part of a domain network you might experience problems with the startup of the PostgreSQL database service In this case it could be necessary to set the password of the Postgres Service manually using the Windows control panel and restart the server again Run the one click installer package to start the installation process Note that you can only use a installer package as an administrator To run the executable with the user rights of an administrator you can right click the executable file and choose run as administrator If you are not allowed to run an executable as an administrator and or you do not have sufficient user rights on your local machine ask your system administrator for support Please follow the instructions and use any given default settings throughout the installation process Note that the installation routine fails to se
106. slave sh The first one starts the agent platform with the single top level controller cf Section 3 1 The top controller must be started first The second command has to be executed once on each additional simulation host The IP address of the computer that executes the top level controller or its name as listed in etc hosts have to be set as the server master IP in the lt plasma_dir gt etc plasma properties file The script starts a remote JADE container and a PlaSMA sub controller which registers itself with the top level controller Linux prompt server slave sh Windows prompt server slave bat When all sub controller have been registered with the top level controller the system is prepared for the execution of a distributed simulation The maximum number of agents which should work on the respective simulation computer is specified in the file lt plasma_dir gt etc plasma properties in key server maxAgents The default value is 1000 and based on experience For other agent implementations it may not fit as well In the end it only matters that server maxAgents properly reflects the difference of computational power and working memory for all computers participating in a distributed simulation The top level controller manages the distribution of the agents among all hosts participating in the simulation A new agent will be started at the computer with the smallest ratio from currently running agents to maximal allowed agent
107. t at a time and would thus have no consistent position Thus the suggested way to handle autonomous entities is to distinguish agents from physical objects and services they provide and exclude them from the ontology if possible 5 4 3 Scenario Ontologies Usually your own scenario consists of at least two or three ontology files the top level domain ontology TLO tlo owl and your scenario specific ontology and possibly other ontologies for sub domains such as transport or production cf Section 5 4 The scenario ontology is supposed to be located in the respective scenario s sub directory owl The predefined ontologies can be found in the PlaSMA sub directory etc owl If you have a adapted copy of the predefined ontologies in the scenario s sub directory owl this is used instead for the scenario As already mentioned you are able to modify predefined ontologies because you can use a local scenario copy but you should not change the top level ontology to avoid that PlaSMA cannot load your scenario The actual scenario ontology particularly contains ABox information such as objects in your scenario You have to create this ontology by yourself Either you use a text editor preferably one that supports XML syntax highlighting or a graphical ontology editor like Prot g Using a Text Editor The following XML document defines a very simple scenario that only uses top level ontology concepts and properties The ontology is na
108. t important ontology concepts for physical world infrastructure modelling 5 4 1 Physical World Infrastructure As stated above the simulation system s physical world model is specified by a directed graph consisting of nodes and edges Nodes or vertices are abstract locations in your world that may correspond to e g traffic junctions as well as logistic sources and or sinks In production scenarios a machine may be modeled as a node as well Edges are directed links between nodes e g roads or railways Nodes are at least typed as tlo Vertex They need to have a location which is stated in geographical coordinates using the datatype properties tlo latitude and tlo longitude Edges have the basic type tlo Edge Each edge must have a specific length stated in kilometers property tlo edgeLength and exactly one start and end location respectively These locations are defined by nodes as values for the properties tlo startingAt start and tlo directedTo end Note that one always has to create two edges to model a link that is trafficable in both directions Types of Nodes and Edges The transportation ontology trans provides several types of nodes and edges The basic edge types are trans LandRoute trans WaterRoute and trans AirRoute trans LandRoute is further divided in trans Road for wheeled or tracked vehicles and trans RailRoad for trains A trans WaterRoute may be either a trans InlandWaterway or a trans SeaWaterway Sub types
109. tallation of Apache Ant on your system Binary distributions 2To use code completion and remove the cannot be resolved to a type errors of Eclipse you might want to add lt plasma_dir gt lib plasma jar lt plasma_dir gt lib jade jar lt plasma_dir gt lib commons jar lt plasma_dir gt lib logger jar lt plasma_dir gt lib worldmodel jar lt plasma_dir gt lib toolkit jar and lt plasma_dir gt lib plasma jcncl jar as external JARs to your project according to your needs This step is optional since the Ant build file has a reference to those JAR files itself 5 4 WORLD MODEL DESIGN 33 are available from the project website for Windows users In most Linux distributions Ant can be installed conveniently via the distributions package management system In order to invoke the Ant build targets for your scenario from the command line switch to the scenario root directory where the build xml file is located Here you tell Ant to invoke either a single target or a comma separated target sequence using the following syntax prompt ant target1 target2 Figure 5 4 shows an example session First the scenario workbench is cleaned Afterward the scenario is re deployed as a multi step process A previous now outdated version of a given scenario is uninstalled from the PlaSMA system Afterwards the scenario sources are compiled put into a Jar archive and finally re deployed i e installed for use in the PlaSMA system
110. the order is type database user address and authentification method Without password authentication host plasma plasma 192 168 100 0 8 trust Same With password authentication host plasma plasma 192 168 100 0 8 md5 Furthermore you have to configure the name resolution of all simulation hosts so that all host names including the own host name and localhost can be looked up On Linux systems this is done in the file etc hosts On windows systems this file is called C Windows system32 drivers etc hosts On the host simulationHost_01 the entries may look as follows 192 168 100 210 simulationHost_01 localhost 192 168 100 211 simulationHost_02 192 168 100 212 simulationHost_03 The configuration of the other hosts is analogous but with localhost at the corresponding line In addition it has to be assured that no firewall is blocking the ports 1098 and above Otherwise the RMI connection of the JADE framework will be blocked Accordingly you have to replace localhost with the host name of the database server in the file lt plasma_dir gt etc serverconfig xml 1 The default location is C Program Files PostgreSQL 9 1 data for Windows or etc postgresql 9 1 main for Linux 3 2 PLASMA CLIENTS 17 Execution For the simulation the same version of the PlaSMA scenarios if used must be present on each simulation host There are two commands available for execution server master sh and server
111. ther things the status bar at the bottom of the window will show the state of the server if connected progress of the simulation time see Section 5 6 and the number of running agents 3 2 2 Command Line Interface There is a command line interface abbr cli without any visualization too It is started with the following command 3Note that your graphics adapter needs to be set to 32 bit colour depth for proper visualisation If you encounter additional problems you should update the driver of the graphics adapter 4See Chapter 4 see Figure 3 4 for a full description of all available scenario options 3 2 PLASMA CLIENTS 19 12 PIESMA Aurora EF Server Options E noO i semver Log WorldWind Visualisation 3 Scenario Configuration WorldWind Layers W PlaSMA Layers I View Configuration Movable Objects 54 Stationary Objects 68 EUROPE Altitude 964km 15 Server State Running Agents 29 Simulation Time 2009 01 0500 18 40 Syncs 7403 Rune 1 1 Progress a 36 Figure 3 2 The WorldWind Visualization Tab shows the infrastructure and all physical objects Drs EF Server Options Ej Select Scenario y gt LL a TB Server Log E WorldWind Visualisation 2 Scenario Configuration Bj Save Save As senio Opton geome Scenario Options Description Default Logistics Scenario Germany Log Level INFO y Re
112. tion attribute and creating an object out of it that should equal other objects created with the same string but in another class or agent It is also recommended to override the hashCode method Though it most likely will not be an issue it could affect the reproducibility by changing the behavior of a HashMap or creation of a random seed And those kind of errors are usually hard to track down 5 6 Time Management and Synchronization Time management and synchronization are an integral part of the PlaSMA simulation system The following text which has been inherited from Geh08 pp 34 provides an introduction with regard to these topics To begin with it is necessary to distinguish different notions of time related to multiagent based simulation MABS Generally physical time refers to the time at which simulated events happen in the real world Simulation time or virtual time models physical time in simulation Wall clock time refers to the time that is consumed by the simulation program executing the simulation As simulation speed might differ between agents each of them has its own local virtual simulation time or in short LVT Jef90 In contrast to equation based modeling cf PSR98 time progression is discrete in MABS Equidistant steps as applied in time stepped simulation are hardly efficient since time steps are even processed if no events occur Fuj00 and inadequate if time steps are too long because events are conside
113. tion of these logical processes LPs In PlaSMA an LP is realized by subclassing of SimulationAgent class SimulationAgent is a base class for implementing agents in PlaSMA It is not necessary to model any ontological counterpart for those simulation agents see Section 5 4 2 or linking it to any ontological instances However the implementation of a simulation agent usually models the control over a specific set of physical objects e g trucks and containers or environmental processes since physical objects are neither given any control nor do they carry a model of technical intelligent behaviour As an exception you could model a robotic agent that is a physical object and a control system at the same time 5 5 1 Agent Setup When subclassing from SimulationAgent one has to override the userSetup method userSetup is being executed when the agent is brought to life It replaces the setup method known from normal JADE agents A PlaSMA simulation agent will already be registered at the agent platform AMS and is therefore capable of sending and receiving messages Anyhow the simulation agent will be registered at one of the PlaSMA sub controllers after userSetup finished Registration to the PlaSMA controller is necessary for simulation time synchronisation and agent life cycle management Thus when userSetup is called the simulation agent is not actually running yet In the userSetup method one can do start up routines li
114. tional multiple Places an image above of the world wind background src required A relative path from the scenario base directory to a local background picture Transparency of PNG files is supported north required The top latitude of the background south required The bottom latitude of the background west required The left longitude of the background east required The right longitude of the background To summarize a visualization configuration might look like the following example lt visualization gt lt images gt lt image key agent1 src images agent1 png gt lt image key http plasma informatik uni bremen de owl test owl Truck src images Truck png gt lt images gt lt worldwind mode sphere gt lt initposition latitude 53 5454 longitude 9 8064 altitude 300000 gt lt background north 55 059 west 5 861 south 47 264 east 15 046 src images germany png gt lt worldwind gt lt visualization gt The image mapping includes an object instance image for agent1 and a general object type image for trucks The initial camera position is focused to Bremen in Germany from an altitude of 300 kilometers Also a background image with the area of Germany was added 4 3 Simulation Logging The logging component of PlaSMA provides the ability to write messages to the database or the console to retrace and analyze a simulation run 4 3 1 Log Levels A log level defines the type and lev
115. ts ConfigAttribute ofType BasicRI class private ConfigAttribute data private CustomAttribute ConfigAttribute data this data data Override public ConfigAttribute data return data Lines 12 17 enable the basic functionality of the ConfigKeyInterface and can be inherited to every ConfigKeyInterface implementation The data object encapsulates parsing and storing of the attribute The first lines 8 10 add enum values and make use of the constructor in line 14 Those values needs to be spelled exactly like the attribute s key in the configuration The ConfigAttribute parameter can be created with the ofType Class lt T gt method If the attribute contains a default value or a list of default values this can be set by chaining the withValues T function It is also possible to add a description by chaining withDescription String Those default values can be used by the scenario configuration editor as suggested value Alternative Attribute Access Although the previously described method is the recommended way because it supports the scenario configuration editor and makes the access to the attributes clearer there is an alternative way to get to the attribute s value All attributes are stored in a Map lt String List lt String gt mapping that maps the attribute s key to its value s in String representation This map can be retrieved by calling the agent s getArguments function and extracting the first object f
116. tup up the database automatically on a file system other than NTFS Therefore we strongly recommend to choose an installation directory that resides in a NTFS file system During the installation process you are prompted for a password for the database superuser namely postgres which is also used as a Windows service account Later you will have to configure PlaSMA with this database user information Now the basic components of PostgreSQL are being installed At the end of the installation process you are being asked whether the Stack Builder application should be run You should request no and manually start Stackbuilder as an Administrator from the Windows start menu The Stack Builder application helps you to install additional components of PostgreSQL Please choose to install the following additional components which are required for PlaSMA when storing simulation data is desired e JDBC database driver pgJDBC in category Database Drivers e ODBC database driver psqiODBC in category Database Drivers e optional PostGIS see 5 4 4 driver PostGIS in category Spatial Extensions Then the selected additional components are being downloaded and installed by the Stack Builder setup wizard Please check during installation of PostGIS whether you specified the installation path of PostgreSQL correctly Please continue with Section 2 1 1 PostgreSQL Configuration for PlaSMA Installation procedure Ubuntu It is recommended to instal
117. uction of the minimal progress can increase the precision of the time laps compared to the real scenario Contrariwise this will decrease run time performance because synchronization overhead increases This attribute uses the same format as the maxSimLength attribute see Section 4 1 maxSyncDist optional Specifies the maximum simulation time that might elapse between two synchronization cycles If this limit is exceeded a synchronization step is forced automatically A fixation of the maximum simulation time progress causes all agents to be waked up in the given interval That can be helpful to guarantee a certain granularity for monitoring and analysis of the scenario This is especially meaningful if the scenario is visualized Furthermore an incorrect programming of the agents can cause that agents are waked up nevermore if unintentionally they receive no further messages An ensured wake up using maxSyncDist can resolve this problem but is not advisable in general This attribute uses the same format as the maxSimLength attribute see Section 4 1 timeout optional Sets the agent response timeout in seconds for synchronization cycles The default value is 5 If an agent computes longer than this time the PlaSMA controller will 1 If an agent is waked up nevermore it has finished its last processing cycle passively That is it has not requested an explicit wake up but is waiting for incoming messages of other agents 4 2 DO
118. usage of it The default value is true Use this to turn off the database without deleting the whole element host optional The address of the database host Can be a plain IP address or a name e g 192 168 0 xx or localhost The default value is localhost name required The name of the database created for PlaSMA user required The username to connect to the database host password required The password credentials to connect with the formerly stated username to the database host dbtype required Currently PlaSMA only supports PostgreSQL databases So the only available value is POSTGRESQL lt console gt optional once This element configures the console output channel If omitted the console output is disabled enabled required Set true to displaying log messages on console colored optional Set true to color the output just on supporting platforms e g Linux The default value is true lt category gt optional multiple The log categories may be hierarchically ordered by providing an absolute path e g name System Communication Input If e g the category Sys tem Communication is not specified it will be implicitly created Additionally it will inherit all settings from the category System The ordering in which these categories are specified has no effect The loglevel attribute is optional if it is not provided it is inherited from its parent category name required Defines the hierarchical l

Download Pdf Manuals

Related Search

Related Contents

Philips PET729  RTMU Control & Monitor Relay User Manual  Orbit 57030 Instructions / Assembly  Manuel d`utilisation LiftNStore Systeme d`entreposage aérien  manual - WorldSpace Foundation  A-dec 200 — Mode d`emploi  Bedienungsanleitung - HENSEL Mess  Furuno FAR-2107(-BB) Marine RADAR User Manual  Origin Storage 500GB  0 - Oldsmobile  

Copyright © All rights reserved. Failed to retrieve file