CupCarbon: A Multi-Agent and Discrete Event Wireless Sensor
Contents
1. network and nodes EURASIP Journal on Embedded Systems C Based Design of Heterogeneous Embedded Systems archive 2008 3 January 2008 T Issariyakul and E Hossain Introduction to network simulator ns2 Book Springer Ed 2011 A K pke M Swigulski K Wessel D Willkomm P K Haneveld T Parker O Visser S L Hermann and S Valentin Simulating wireless and mobile networks in omnet 4 The mixim vision In the 1st International Conference on Simulation Tools and Techniques for Communications SIMU Tools 08 Marseille France March 3 7 2008 K W Lee and W K D Sensim A simulation program for solid state pressure sensors In the IEEE Transactions on Electron Devices 29 1 34 41 August 9 2005 P Levis N Lee M Welsh and D Culler Tossim Accurate and scalable simulation of entire tinyos applications In the 1st International Conference on Embedded Networked Sensor Systems pages 126 137 New York USA 2003 P Pagano M Chitnis and G Lipari Rtns an ns 2 extension to simulate wireless real time distributed systems for structured topologies In the 3rd International Conference on Wireless Internet WICON 07 Austin USA October 22 24 2007 A Rastegarnia and V Solouk Performance evaluation of castalia wireless sensor network simulator In the 34th International Conference on Telecommunications and Signal Processing TSP pages 111 115 Budapest Hungary August 18 20 2011 A Sethi J P Sain2. these agents Each sensor agent executes its associated communication script after its initialization step and begins to move using its associated gps file The script allows the generation of events send wait etc which are recovered by the sched uler to execute them in the order of their creation dates The amm EE Device 1 i 1 DeviceWithoutRadio 1 i 1 FH 1 1 1 1 1 4 eT Ee ee User Interface 1 Devices amp Events OpenStreetMap 1 Database DUUM UE WiSeN Simulator Solver Optimization algorithms Couverage Routing Event List Figure 6 The CupCarbon Architecture Agent initialization Load communication script and gps file Start the agent Simulation time finished Update result file Exclude the agent from the simulation Generate Update an event result file Lock the Unlock the agent scheduler Figure 7 The behaviour of a simulated agent BREAK COM process describing the behaviour of an agent is described in Figure 7 During a simulation an agent starts to generate events de scribed in the associated communication script and those re lated to mobility These will then be communicated to the scheduler int order to sort them for a next processing phase As long as the agent has not met the stopping conditions it generates one event at a time wheneve
3. which is connected to the input pin 12 of the card and then will send this read value wirelessly int p 12 void setup pinMode p INPUT Serial begin 9600 void loop function delay 1000 void function Serial write 134 int x digitalRead p Serial write x Figure 1 Example of Arduino program A Wireless Sensor Network is an ad hoc network and is composed of a number of sensor nodes that are able to col lect send and receive autonomously environmental data via wireless communications Figure2 shows an example of a WSN The position of these nodes is not necessarily prede termined They can be dispersed randomly in a geographical area called Wellfield corresponding to the field of interest Base Station Sensors mobile 25 gt a Jag 9 5 A Date Processing LM Base Station static Figure 2 Example of a Wireless Sensor Network WSN is one of the most important technologies that have changed the world and facilitate many daily tasks These WSNs offer the possibility of observing and controlling phys ical and biological phenomena in several areas industrial scientific temperature pressure humidity light etc the environment pollution CO2 etc health patient mon itoring epidemiological studies etc transport accident prevention etc home automation and so on After being captured the data is routed based on a multi hop
4. routing to a node which is considered as a collection point called Base Station The Base Station depending on its configuration can store process or transmit the received data and it can be connected to the network via internet satellite or any other system The user can send requests to the nodes of the network specifying the type of data needed and collect environmental data from the base station The data routing is a critical phase for the lifetime of the network as the routing depends on the energy state of the sensors in the network and on the data constraints 3 THE CUPCARBON SIMULATOR CupCarbon is a multi agent and discrete event wireless sen sor network simulator which is based on geolocation It al lows to model and simulate sensor networks on a digital geographic interface of OpenStreetMap For this purpose CupCarbon provides a set of easy to use and configurable objects Figure 3 shows the graphical interface of this sim ulator The use of multi agent systems allows better opti misation of the simulation time by taking advantage of the parallelism of agents and events C upCarbon is composed of three main components a multi agent simulation envi ronment mobile simulation and finally the WSN simulator WiSen These three components will now be described 3 1 Multi agent Simulation Environment Cup Carbon provides an environment for a multi agent simu lation which allows to run simulations and monitor var
5. 95910644531 0 25 Figure 4 Example of a GPS file HEAD PROTOCOL PROTOCOL ID send 154 70 S2 send 154 100 3 S4 send 154 90 S1 S8 S9 wait 200 send 154 90 S2 3 S6 send 154 30 wait 500 send 154 50 S1 S8 S9 break Figure 5 Example of a communication script GPS file is considered as a static agent and a GPS file can be assigned to multiple agents 3 3 Communication script CupCarbon allows to configure each agent of the simulated system using script files to program their communications Such a file describes how an agent will communicate with its neighbours and its environment It contains all the main actions to be performed by a sensor during the simulation Each line of the file represents an event that is characterised by its type date value signal power and the list of agents involved in it An event is a command that can be of five different types as described below e COM_SEND for the packet message sending e COM_DELAY COM_WAIT to force a waiting e COM_BREAK to mark the end of a communication e COM_RECEIVE to accept the reception of a packet mes sage e COM_UNKNOWN for unknown commands This file can in some cases have a header part which allows to configure the communication protocol and describe the type of algorithm to use in the simulation subject to the fact that the algorithm is already integrated into the simulator The header commands are given in the f
6. ResearchGate See discussions stats and author profiles for this publication at http www researchgate net publication 2608 71305 CupCarbon A Multi Agent and Discrete Event Wireless Sensor Network Design and Simulation Tool ARTICLE MARCH 2014 DOI 10 4108 icst simutools 2014 254811 DOWNLOADS 100 4 AUTHORS Kamal Mehdi University College Dublin 2 PUBLICATIONS O CITATIONS SEE PROFILE Ahcene Bounceur Universit de Bretagne Occidentale 41 PUBLICATIONS 177 CITATIONS SEE PROFILE VIEWS 18 Massinissa Lounis Universit de B ja a 3 PUBLICATIONS O CITATIONS SEE PROFILE Tahar Kechadi University College Dublin 214 PUBLICATIONS 682 CITATIONS SEE PROFILE Available from Ahcene Bounceur Retrieved on 16 June 2015 CupCarbon A Multi Agent and Discrete Event Wireless Sensor Network Design and Simulation Tool Kamal Mehdi School of Computer Science amp Informatics University College of Dublin Belfield Dublin 4 Ireland Kamal Mehdi ucdconnect ie Massinissa Lounis LIMED Laboratory University of Bejaia Bejaia Algeria Massinissa Lounis univ bejaia dz Ahcene Bounceur Lab STICC Laboratory Universiy of Brest Brest France Ahcene Bounceur univ brest fr Tahar Kechadi School of Computer Science amp Informatics University College of Dublin Belfield Dublin 4 Ireland Tahar Kechadi ucd ie ABSTRACT This paper presents the first version of a Wireless Sensor Network simul
7. Sensim 9 which allows to develop new protocols and test their scala bility The third family of simulators is based on Ptolemy IT 1 which is a framework for modeling simulating and designing of parallel embedded real time systems We can mention two simulators of this family Viptos 5 integrating graphical development framework to simulate WSN based on TinyOS and VisualSense 2 which can be used as an educational tool to understand the basics of WSNs Unfor tunately these two simulators are not available for download The fourth family is specially developed for WSNs such as WSNet 4 used to evaluate high level designs Atarraya 16 for teaching and research topology control algorithms and J Sim 10 a framework for modeling and simulation In the context of this study we present a simulator named Cup Carbon which is based on multi agent and discrete event simulation The current version belongs to the second fam ily of simulators that is described above It can be used to generate networks for OMNET 4 It offers a simple and friendly graphical user interface for the modeling of the net works using the OpenStreetMap OSM framework Each node is designed to be as close as possible to the real one It is composed of four modules mirco controller the radio an tenna the capture unit and a battery The current version of CupCarbon includes simulation of mobiles and it allows to represent the detailed energy diagram for each no
8. ator called CupCarbon It is a multi agent and discrete event Wireless Sensor Network WSN simula tor Networks can be designed and prototyped in an er gonomic user friendly interface using the OpenStreetMap OSM framework by deploying sensors directly on the map It can be used to study the behaviour of a network and its costs The main objectives of CupCarbon are both educa tional and scientific It can help trainers to explain the basic concepts and how sensor networks work and it can help sci entists to test their wireless topologies protocols etc The current version can be used only to study the power diagram of each sensor and the overall network The power diagrams can be calculated and displayed as a function of the simu lated time Prototyping networks is more realistic compared to existing simulators Keywords Wireless Sensor Network simulator multi agent system dis crete event simulation OpenStreetMap mobility 1 INTRODUCTION Progress and development in the field of wireless commu nication and electronics in recent years have given rise to a new technology known as Wireless Sensor Networks WSN These are composed of a large number of sensors that can be deployed randomly and densely A sensor is a small elec tronic device that can collect data from its environment and send it to a base station The type of data collected varies depending on the application and the type of the sensors Sensor networks have man
9. critical resource The scheduler repeats this cycle until the stopping conditions are met The overall simulation stops when one of the following three conditions is fulfilled e there is no events to execute in the scheduler list e the simulation time is exceeded e there is a forced stop or real time limitation of the simulation 4 CASE STUDY In this section we will show how to use C upCarbon to sim ulate the energy diagram of a sensor network which is de signed on the OpenStreetMap 4 1 Simulation Setup Once the CupCarbon tool is started and a new project is created it is possible to create a network of sensors directly on a geographical map which represents the main window of the software It is possible to create routes to be assigned to mobiles It is also possible to create mobiles gas and a cloud of insects For the sake of simplicity we have chosen as an example a simple network consisting of eight sensors as shown in Figure 9 900 Figure 9 A WSN designed with CupCarbon Once the network is developed the different communication scripts must be created and assigned to each sensor One script can be assigned to multiple sensors In our case we create a single script that is assigned to all sensors The script is shown in Figure 10 This script allows a sensor to send 10 bytes with the maximum signal power 100 send 100 100 send 110 100 send 120 100 send 130 100 s
10. de versus the simulation time This version can be also considered as a kernel which can be used to integrate different algorithms and modules making use of its advantages It can simulate mobile tracking scenarios This paper is organised as follows Sensor networks will be introduced in the Section 2 Section 3 will present the CupCarbon design and simulation tool Some simulation results will be presented in Section 4 The conclusion will be given in Section 5 2 WIRELESS SENSOR NETWORKS Sensors are electronic components that operate in networks with autonomy The term sensor is usually used to refer to a sensor node having a dimensions of a few centimeters It is composed of four components which are a microcon troller programmable integrated circuit a radio antenna for wireless communication a battery and a set of sensors that we will call capture unit in order to avoid confusion A capture unit having a dimension of a few millimeters cap tures or intercept environmental information motion tem perature humidity gas etc Henceforth we use the term sensor to refer to a sensor node and sensor unit to refer to a sensor In order to be able to communicate read or receive infor mation a sensor must be programmed Figure 1 shows an example of a program for Arduino cards with an AT mega microcontroller This program will send each second the byte 134 and will read the value of a sensor unit motion sensor
11. e structure of the network We have shown that a simple delay of only 100 milli seconds in the communication script of a sensor can improve significantly its lifetime 6 REFERENCES 1 P Baldwin S Kohli E A Lee X Liu and Y Zhao Modeling of sensor nets in ptolemy ii In the 3rd International Symposium on Information Processing in Sensor Networks IPSN 04 Berkeley USA pages 359 368 April 26 27 2004 2 P Baldwin S Kohli E A Lee X Liu and Y Zhao Visualsense Visual modeling for wireless and sensor network systems Technical Memorandum UCB ERL M04 06 University of California Berkeley USA April 23 2004 3 A Boulis Castalia a simulator for wireless sensor networks and body area networks Users s Manual version 3 2 March 2011 4 11 12 G Chelius A Fraboulet and E Fleury Worldsens development and prototyping tools for application specific wireless sensors networks In the 6th International Conference on Information Processing in Sensor Networks pages 176 185 Cambridge USA April 25 27 2007 E Cheong E A Lee and Y Zhao Viptos A graphical development and simulation environment for tinyos based wireless sensor networks Demo Abstract in the 38rd ACM Conference on Embedded Networked Sensor Systems SenSys 05 page 302 San Diego USA November 2 4 2005 J Glaser D Weber S A Madani and S Mahlknecht Power aware simulator framework for wireless sensors
12. end 140 100 send 150 100 send 160 100 send 170 100 send 180 100 send 190 100 delay 100 Figure 10 Communication script of the sensors As soon as all communication scripts are associated with the sensors the simulation can begin The next section describes the results for the network presented above 4 2 Simulation Results Once the network is ready the real simulation parameters must be specified the simulation time and the simulation step Thus the simulation can begin In our case we chose to simulate a period of 100 hours 360K seconds with a simulation step of one hour Indeed this parameters are given by the user but the execution will be based on the discrete event simulation Figure 11 shows the curves representing the energy diagrams obtained for each sensor In this graph eight curves are drawn but we distinguish only four because the other curves are superimposed In this example these curves are directly related to the number of neighbours of each sensor There are four types of sensors corresponding to the four families of curves We distinguish the family of sensors with only one neighbour sensors S6 and 7 in blue the sensors with two neighbours sensors S1 S2 and S4 in red the sensors with three neighbours sensors S5 and S8 in purple and only one sensor with four neighbours S3 sensor in green As we can see on the graph for the script of the Figure 10 wi
13. esponds to an event One log file is generated per simulation 2 rst file a result file generated for each sensor It contains the energy evolution of a sensor during the simulation 3 5 CupCarbon Architecture CupCarbon is developed in Java Its architecture consists of two layers The first layer concerns the modules used to build the simulation The second layer concerns the simula tion itself Figure 6 shows the different modules of CupCar bon Four main modules can be distinguished e Agents Module it includes devices and events nec essary to prototype wireless sensors networks and to prepare and configure the simulator e OpenStreetMap Module it allows designing wireless sensors on an OSM map e WiSen Simulator Module it allows to simulate wire less sensor networks This module is connected to the simulator agents e Solver Module it integrates a set of optimisation al gorithms such as routing coverage etc The current version integrates an algorithm for solving the set cov ering problem used to determine the minimum number of sensors that detect the maximum of targets This part will not be presented because it is not a part of the objectives of this paper In order to simulate wireless sensor networks two main types of actors are needed the agents and the scheduler Once cre ated and initialised the simulator then starts the agents and the scheduler and synchronises all critical sections between
14. i and M Bisht Wireless adhoc network simulators Analysis of characterstic features scalability effectiveness and limitations International Journal of Applied Information Systems IJAIS 5 9 July 2012 A Varga Omnet User s Manual version 4 83 A Varga and R Hornig An overview of the omnet simulation environment In the 1st International Conference on Simulation Tools and Techniques for Communications SIMUTools 08 Marseille France March 3 7 2008 P M Wightman and M A Labrador Atarraya A simulation tool to teach and research topology control algorithms for wireless sensor networks In the 2nd International Conference on Simulation Tools and Techniques for Communications SIMU Tools 09 Rome Italy March 2 6 2009
15. ious CupCarbon Pee Ag pay i Piae t Pu gt t Met aay o T m j TR E Markers gt 3 w o i 7 mobile Oute A d ES 4 s Mobile sensor 9 ead Quum d 208 EIL t j Ae fez rd B E P Sensor parameters moss ursa Figure 3 The Interface of CupCarbon events and changes over time The use of a multi agent sys tem allows a reproduction of an environment similar to the real world system wherein each object in the system oper ates autonomously while communicating with other objects in the same environment The integration of OpenStreetMap in this environment provides an interface and a database of digitised data related to geolocation such as roads the po sitions of buildings etc Among the main agents in the sys tem one can find sensors for sending and receiving packets There is also gas insects and mobiles They do not neces sarily communicate between them but they interact with the sensors In the next versions the agents will be able to in teract with the environment like buildings and meterological events e g wind temperature etc 3 2 Mobility Cup Carbon offers the possibility to create paths free or re lated to real roads and assign them agents to make them mobile Each path is defined by a set of referencing points on the OSM map and each point is associated to a date corre sponding to the exact arrival date
16. nd it dies after only 57 hours with out this delay 000 s z Me Graph Viewer Sensors Energy 000 Ee 95 000 e 90 000 S s3 85 000 Vj s4 80 000 Ass 75 000 j V s6 70 000 V s7 65 0001 60 000 1 55 000 1 gy 50 000 45 000 40 000 1 35 000 30 000 25 000 20 000 15 000 10 000 1 5 000 1 0 Ener 0 50000000 100000000 150000000 200000000 250000000 300000000 350 000 000 Time Si S2 S3 S4 S5 S6 S7 S8 Figure 12 Energy diagrams related to sensors with the delay instruction 5 CONCLUSION A simulator Cup Carbon for the design and study of wireless sensor networks is presented It allows to design networks using the OpenStreetMap framework Networks can include sensors and other elements such as mobiles gas fires etc The presented tool can run two types of simulations The Multi agent simulation is used to parallelise the behaviour of each sensor to make it independent The discrete event simulation is used to simulate the communications between agents and especially between sensors The main objective of this simulator is educational It will demonstrate the use of wireless sensors under almost the same conditions as in the real world The simulator can also be used to calculate the energy diagram of each sensor A case study showing how to obtain this type of diagrams is also presented The results are consistent with th
17. of the object to it There are two ways to generate trajectories in this environment 1 Manual generation it is possible to generate trajec tories by manually selecting the points x y on the map and by specifying the arrival dates at each point 2 Automatic generation it is also possible to automat ically generate trajectories on the OSM map by speci fying the main points of the trajectory Paths must be saved For memory optimisation reasons only trajectories associated with objects included in the sim ulation are loaded into memory Only one point is loaded in each move The coordinates x y of given objects will be updated at the specific date for each point in the path during the simulation One trajectory can be associated to multiple objects of different types It is represented by a GPS file where each line describes a date GPS coordinates a boolean parameter and the radius of the point to draw on the map The boolean parameter is used to determine whether to draw the mobile objects connected at this point or just simulate without drawing It differentiate a simu lation for visualisation purposes and the simulation for the calculating the energy diagram Figure 4 shows an example of a GPS file Note that an agent who is not linked to a 00 00 00 36 787291466820015 4 965476989746094 0 25 00 00 01 36 768042206102585 4 956550598144531 0 25 00 00 02 36 74906320434928 4 95208740234375 1 25 00 00 03 36 73833386539537 5 0004
18. ollowing e COM_HEADER indicates the inclusion of a header e COM_PROTOCOL this is for the communication protocol to be used in the simulations Note that the headers are optional in the communication script files and in the case of the absence of the header a standard protocol will be used Also in the absence of the break command the execution of the file will be done repeatedly until the end of the simulation Figure 5 shows an example of a communication script file 3 4 The WSN simulator WiSeN WiSeN is the name of the module that represents the kernel of CupCarbon for simulating events related to sensors send ing receiving waiting etc It supports and manages the evolution of the state of each object in the system energy position etc Its implementation in a multi agent envi ronment allows for each agent to be executed independently and in parallel Thus it is possible to include mobility as pects and the detection of the target Each agent generates events based on a existing saved script The simulator or ganises events generated by agents sensors mobiles etc according to their creation dates and executes them in the same order and it updates their status energy position etc Two types of files are automatically generated during the simulation 1 log file a file in which all the events executed by the simulator are stored for more detailed analysis and debugging purposes Each line in the file corr
19. r the scheduler unlocks it by calling the synchronisation semaphore The agent un locks the scheduler after generating the event It then enters into idle state until the next call of the scheduler The scheduler is an independent module that runs in parallel with the agents It organises events generated by agents in order to execute them It is composed of a list of events a management algorithm and processing Figure 8 shows the main role of the scheduler The scheduler is started just after its initialisation Then it will run in the background by collecting at each iteration the next events If the agent of generating events has the necessary energy then the event will be executed otherwise the scheduler move to the next event of the event list After its execution it updates all the agents associated to the event such as receiving a message from a sensor and then add this action to the log file The scheduler agent indicates the event to the generator agent so that it can generate and Scheduler initialization Start the Scheduler Stop condition End of Lock the Unlock the simulation scheduler agent NO Go to the next event 7 Agent Execute Update the Duas the event log file YES empty g Figure 8 The scheduler add the next event in the scheduler list Then it switches to the idle state while the list is used by the agent the list of events is a
20. thout the last line delay and for a given capacity of the battery the sensor S3 having four neighbours in green curve dies after 57 hours 205 200 seconds Note that the selected script means that the sensors send and receive messages continuously while the battery is charged nd 7 En NY NY X NE al Wee E DC 100 mr EIUS e000 Graph Viewer Sensors Energy E 95 000 90 000 85 000 80 000 jf 75 000 70 000 65 000 60 000 55 000 gt amp 50 000 45 000 40 000 35 000 30 000 25 000 20 000 15 000 Green Purple LO OOD ees nena ence aaa ting nacanrnenenecattanaenesneneamenreenesnatenmtta 5 DOD 9n nmn E 50 000 000 100000000 150000000 200000000 250000000 300000000 350000000 Time S1 S2 S3 S4 S5 S6 S7 S8 S Rie Ke Figure 11 Energy diagrams related to sensors without the delay instruction If we take into account the delay of 100 milli seconds for each outgoing 10 bytes as shown in the last line of the script of Figure 10 we can improve significantly the lifetime of the sensors This new situation is illustrated by the curves in Figure 12 After 100 hours functioning the battery of the sensor S3 reaches 70 of its total capacity In other words the sensor S3 consumes 30 of its battery each 100 hours functioning with an additional delay of 100 milli seconds in its communication script a
21. y applications in different fields such as health environment agriculture geology military etc Most applications of WSN are challenging for designers and this is due to the limited capacity of the nodes in terms of autonomy battery computing power and inaccessible areas This makes the design of algorithms and programs for WSN too constrained Therefore performance evalua tion tools become very essential in the process of designing a wireless sensor network and simulation is one of the most used tools for this evaluation There exists a large number of WSN simulator tools Some of them are described in 12 In this work we propose to separate these tools into two main categories simulators and emulators and to introduce only some simulators that are of four different types The first type represents simu lators that are based on NS2 7 which is a simulator de veloped in general for traditional networks and adapted for WSN It uses discrete event simulation In this family we can find NRL Sensorsim 13 based on modules and RT NS 11 for the real time distributed systems The second fam ily is based on the OMNET simulator 14 It is a dis crete event simulator We can find Castalia 3 for the de velopment of protocols and distributed algorithms MIXIM 8 an inter level platform NesC T 15 which allows to run TinyOS applications Pawis 6 with its modular design al lows to simulate deferent types of nodes and
Download Pdf Manuals
Related Search