1. VANET:On mobility Scenarios and Urban Infrastructure. & Realistic Simulation of Network Protocols in VANET Scenarios Advisor: Kai-Wei Ke Speaker: Chia-Ho Chao Date: 22/04/2008 2008/4/221

2. VANET:On mobility Scenarios and Urban Infrastructure. 2008/4/222

3. Outline Overview of TRANSIMS Overview of CORSIM Random waypoint (RWP) Mobility Models Comparison Flat Network Opportunistic Infrastructure Inter-contact Times Afternoon Trend Conclusion 2008/4/223

4. TRANSIMS Traces TRANSIMS : Transportation Analysis Simulation System Asses the performance of a large scale urban sensor network. Creating car movement patterns based on activity flows by large scale, vehicular traffic and parallel simulator. 2008/4/224

5. CORSIM Traces CORSIM : Microscopic Traffic Simulation Model high level of precision in vehicular traffic simulation. difference from TRANSIMS: Single CPU Lack of activity flow information 2008/4/225

6. RWP 2008/4/226

7. Mobility Models Comparison: Simulation setting 2008/4/227

8. Simulation setting VANET simulations are run for 200 seconds on a 1*2 km rectangle on the map. Highest AP density 7AM8AM Average vehicles270371 Average speed per second 12.6 m/s 12.5 m/s Stop time(seconds)3.25.7

9. Mobility Models Comparison: FLAT NETWORK 2008/4/229

10. Using TRANSIMS mobility traces 2008/4/2210 8am no APs 7am no APs

11. Using RWP model 2008/4/2211 8am no APs 7am no APs

12. Using CORSIM mobility traces 2008/4/2212 8am no APs 7am no APs

13. Mobility Models Comparison: OPPORTUNISTIC INFRASTRUCTURE 2008/4/2213

14. Simulation Setting 2008/4/2214

15. Simulation Setting 2008/4/2215

16. Using TRANSIMS mobility traces 2008/4/2216 8am with APs 7am with APs

17. Using RWP model 2008/4/2217 8am with APs 7am with APs

18. Using CORSIM mobility traces 2008/4/2218 8am with APs 7am with APs

19. Mobility Models Comparison: Inter-contact Times and Afternoon trend 2008/4/2219

20. TRANSIMS mobility traces at 7AM 2008/4/2220

21. TRANSIMS mobility traces at 8AM 2008/4/2221

22. Afternoon trend 2008/4/2222

23. Afternoon trend 2008/4/2223

24. Conclusion The results confirm that open APs can be effectively exploited to dramatically improve performance. A correct model of traffic flows is important. In long timeframes during the day (i.e. rush hours), network becomes static. 2008/4/2224

25. Realistic Simulation of Network Protocols in VANET Scenarios 2008/4/2225

26. Outline Traffic Simulation Network Simulation Coupling Traffic Microsimulation and Network Simulation Simulation Result Conclusion 2008/4/2226

27. Traffic Simulation Macroscopic models METACOR Mesoscopic models CONTRAM Microscopic models Cellular Automaton model (CA) SK model IDM/MOBIL model 2008/4/2227

28. Intelligent-Driver Model(IDM) Car-following model 2008/4/2228 The gap to a vehicle in front Difference in speed Desired velocity Time headway a:comfortable acceleration b:comfortable deceleration Acceleratio n exponent Additional gap(driving) Minimum gap(jam) Desired gap acceleration

29. MOBIL MOBIL: Minimizing Overall Braking decelerations Induced by Lane change Have to be fulfilled two criteria: a)The lane change has to be safe. 2008/4/2229 Maximum safe deceleration Desired gap Acceleration b) Bias to the right lane politeness factor Lane change threshold

30. Road Traffic Microsimulation Parameters CarTrack Desired velocityV0V0 33.0m/s22.2 m/s Time headwayT1.5s1.7s Comfortable accelerationa0.73m/s^2 Comfortable decelerationb1.67m/s^2 Acceleration exponent44 Minimum gap (jam) S 0 2m Additional gap (driving) S 1 0m Vehicle length l6m10m Politeness factor P20% Maximum safe decelerationb save 4m/s^2 Lane change thresholda thr 0.3 m/s^20.2 m/s^2 Bias to the right lane △b△b 0.1 rm/s^20.3 m/s^2 2008/4/2230 δ

31. OMNeT++ A discrete event simulation environment. primary application area is the simulation of communication networks GUI support INET Framework 2008/4/2231

32. DYMO Routing Protocol DYMO: Dynamic MANET On-Demand Reactive 2008/4/2232 ABCD AODV AAA DDD DYMO ABABC DC DCB

33. DYMO and support modules in the protocol stack 2008/4/2233 App1 App2 DYMO Transport Layer Network Layer Data Link Layer queue hook TCP.mss1024Byte TCP.advertisedWindow14336Byte TCP.tcpAlgorithmClassTCPReno ARP.retryTimeout1s ARP.retryCount3 ARP.cacheTimeout100s mac.addressauto mac.bitrate11Mbit/s mac.broadcastBackoff31slots mac.QueueSize14Pckts mac.rtsCtsFalse

34. Simulated VANET Scenario 2008/4/2234

35. Coupling Traffic Microsimulation and Network Simulation 2008/4/2235 Car ; i=car0_[…]Car ; i=car1_[…]Car ; I=car1_[…] 3187,1487 3187,1490 3187,1493 3187,1495 3187,1498 3187,1501 3188,1504 […] 3171,1380 3172,1387 3173,1394 3174,1402 3175,1409 3176,1417 3177,1425 […] 3154,1262 3156,1269 3153,1277 3155,1284 3156,1291 3158,1299 3159,1306 […] Excerpt from the traffic simulation’s output stream

36. Simulation Result 2008/4/2236 15%

37. Simulation Result 2008/4/2237

38. Simulation Result 2008/4/2238

39. Conclusion Integrated the traffic model in network simulation in order to improve the quality of network simulations. Simulation setups using simple mobility models often produce skewed results compared to the application of traffic models. 2008/4/2239

40. Reference G. Marfia, G. Pau, E. De Sena, E. Giordano, M. Gerla, “Evaluating Vehicle Network Strategies for Downtown Portland: opportunistic infrastructure and the importance of realistic mobility models,” http://transims.tsasa.lanl.gov/http://transims.tsasa.lanl.gov/. http://mctrans.ce.ufl.edu/featured/TSIS/Version5/corsi m.htmhttp://mctrans.ce.ufl.edu/featured/TSIS/Version5/corsi m.htm. http://www.omnetpp.org/ I. Dietrich, C. Sommer, and F. Dressler, "Simulating DYMO in OMNeT++," University of Erlangen, Dept. of Computer Science 7, Technical Report 01/07, April 2007. 2008/4/2240

41. 2008/4/2241 Thanks for your attention!