Performance Evaluation of IEEE 802.11p for Vehicular Communication Networks A. Jafari, S. Al-Khayatt and A. Dogman Faculty of Art, Computing, Engineering and Sciences, Sheffield Hallam University, Sheffield, United Kingdom Email: Amir.Jafari@student.shu.ac.uk; email@example.com; firstname.lastname@example.orgAbstract— IEEE 802.11p is an emerging standard which vehicles. Due to the characteristics of VANET and limitedprovides vehicular safety communication through wireless bandwidth, periodic broadcast messages can consume thenetworks. In this paper, the architecture of Wireless Access entire available bandwidth. Furthermore; the emergencyfor Vehicular Environment (WAVE) and IEEE 802.11p messages need to be disseminated quickly and efficiently.standard were analysed. The key parameters of this Consequently, there is a need to prioritise important andstandard are implemented in ns-2 network simulator to time-critical messages and use quality of services. Theaccurately simulate vehicular ad hoc networks (VANETs). IEEE 802.11p MAC layer implements a priority schemeThe performance of this standard was measured in ns-2 in a similar way to IEEE 802.11e EDCA function.network simulation environment using realistic vehicular The contribution of this paper is to evaluate the IEEEmobility models. The main performance metrics for 802.11p standard. A study was based on the structure ofvehicular safety communication; Throughput, End-to-End the WAVE architecture for VANETs. We, subsequently,delay, and Packet loss ratio were analysed for our scenario. set up one real scenario which assisted us in analysing theIn addition, the effect of varying vehicle speed and differentmessage sizes on the performance metrics were measured. performance metrics of the IEEE 802.11p. This scenario was implemented and modelled using ns-2 network I. INTRODUCTION simulator  with VanetMobiSim traffic simulator . One of the most important points in the vehicular network Intelligent Transportation System (ITS) is one of the simulation was that the nature of vehicular communicationinformation and communication technologies which has is based on the movement. Therefore, it is necessary toattracted a lot of attention recently. This technology implement a realistic vehicular movement in theenhances transportation safety, reliability, security and simulation. The main novelty of this paper is to implementproductivity by integrating with existing technologies. the key parameters of 802.11p standard in ns-2, andWireless data communication between vehicles is one of prepare the realistic vehicular mobility model bythe technologies which has improved the deployment of VanetMobiSim. In other words, all of the importantITS applications. This communication is divided into two parameters are implemented accurately in the VANETtypes: Vehicle to Vehicle (V2V) and Vehicle to simulation.Infrastructure (V2I). Vehicles are equipped with shortrange wireless communication technology (approximately Several publications , ,  have studied the100 to 300 metres) acting as computer nodes on the road. performance of 802.11p. However, none of the previousThis is known as vehicular ad hoc network (VANET) studies have supported realistic vehicular mobilitytechnology. The major objectives of VANET technology simulation. In , the authors have presented acan be stated as follows: broadcasts warning messages to comprehensive evaluation and review of the performanceneighbouring vehicles in case of car accidents help of 802.11p and WAVE protocols supporting realisticemergency vehicles to pass other vehicles quickly, and vehicular mobility model. However, standards wereprovides drivers with latest real-time traffic information. implemented in Qualnet network simulator. In terms of modelling accuracy, a new model of IEEE 802.11 MAC A wide range of project activities have initiated around and PHY, which support IEEE 802.11P, is designed andthe world in order to improve vehicular communication implemented in ns-2 network simulator version 2.34 .networks. In 2004, IEEE 802.11 task group p developed This version of ns-2 network simulator is used in thisan amendment to the 802.11 standard in order to enhance paper.the 802.11 to support VANETs. This standard is known as802.11p, it defines physical and medium access control The remainder of this paper is organised as follows. Inlayers of VANETs. In addition, The IEEE 1609 working section II the WAVE and IEEE802.11p structure aregroup defined IEEE 1609 protocol family which clarified. The simulation scenario is conducted in sectiondeveloped higher layer specification based on 802.11p. III. Results from the simulation and the analyses of theThis protocol consists of four documents: IEEE 1609.1, performance metrics of the IEEE 802.11p are presented inIEEE 1609.2, IEEE 1609.3, and IEEE 1609.4. IEEE 1609 section IV. Finally, this paper is concluded in Section V.protocol family and 802.11p together are called WAVE II. VEHICULAR COMMUNICATION BASED ON THEstandard. This system architecture is used for automotive IEEE 802.11P AND WAVE SYSTEMwireless communications . The specific nature of VANET makes it different from In this section we briefly present an outline of WAVEother kinds of networks; some of the characteristics of architecture system and IEEE 802.11p protocol forVANET are high mobility, short communication periods, VANET.dynamic topology and limited bandwidth. Communicationin VANETs is based on event-driven messages orbroadcast messages exchanged between surrounding
A. Physical and MAC Layers The contention procedure between channels to The physical and MAC layers of WAVE are based on access the medium supported by different timerIEEE 802.11p standard. The physical layer of IEEE settings based on the internal contention802.11p consists of seven channels in 5.9GHz band which procedure. is similar to IEEE 802.11a design, but the main difference Logical link control (LLC) is another element ofis that the IEEE 802.11p uses 10MHZ bandwidth for each WAVE structure which is similar to upper sub-layer ofchannel instead of 20MHZ bandwidth in IEEE 802.11a. OSI layer two. LLC provides the communication betweenThe physical layer of 802.11p uses OFDM technology in upper layers and the lower layer.order to increase data transmission rate and overcomesignal fading in wireless communication. One of the C. Network and Transport Layersspecifications of IEEE 802.11p is that the management The IEEE 1609.3 defines the operation of services atfunctions are connected to the physical and MAC layers network and transport layers. Moreover, it providescalled physical layer management entity (PLME) and wireless connectivity between vehicles, and vehicles toMAC layer management entity (MLME), respectively . roadside devices. The functions of the WAVE networkThe IEEE 802.11p uses CSMA/CA to reduce collisions services can be separated into two sets:and provide fair access to the channel. Data-plane services: They transmit network traffics and support IPV6 and WSMP protocols. Resource Manager Security WAVE short-message- Protocol (WSMP) IEEE 1609.1 Services provides this capability that applications can send UDP/TCP WSMP IEEE IPV6 WME 1609.2 short message to increase the probability of LLC IEEE receiving the messages in time. Multi Channel Operation 1609.3 Management-plane services: Their functions are IEEE 1609.4 to configure and maintain system, for instance: WAVE MAC MLME IPV6 configuration, channel usage monitoring, IEEE 802.11p and application registration. This service is WAVE PHY PLME IEEE 802.11p known as WAVE management entity (WME). Figure 1. WAVE Architecture D. Resource ManagerB. Multichannel Operation IEEE 1609.1 standard defines a WAVE application known as resource manager (RM) which allows IEEE 1609.4 is one of the standards of the IEEE 1609 communication between applications runs on Roadsideprotocol family, which manages channel coordination and units (RSU) and On-board units (OBU). The RM residessupports MAC service data unit delivery. This standard on either OBUs or RSUs .describes seven different channels with different featuresand usages (six service channels and one control channel). E. Security ServicesIn addition, these channels use different frequencies and The IEEE 1609.2 standard defines security services fortransmit powers. Eichler  mentions that each station the WAVE architecture and the applications which runcontinuously alternates between the control channel andone of the service channels; however the different through this architecture. This standard defines the formatchannels cannot be used at the same time. According to and the processing of secure messages ., the control channel is used for system control andsafety data transmission. III. SIMULATION The IEEE 802.11p MAC layer is based on multichannel Implementing and deploying VANETs in a real worldoperation of WAVE architecture and 802.11e EDCA. can be prohibitively expensive and difficult.EDCA mechanism defines four different access categories Consequently, most of the researches in the area of(AC) for each channel. The access categories are indicated Vehicular communication network are based onby AC0-AC3, and each of them has an independent queue simulation for evaluation .. The EDCA mechanism provides prioritization by Simulation in VANET consists of two components:assigning different contention parameters to each access traffic simulation and network simulation. Trafficcategory. AC3 has the highest priority to access medium, simulation focuses on vehicular mobility and it generateswhereas AC0 has the lowest priority. Each frame is a trace file which provides realistic vehicles movement.categorized into different access categories, depending on This trace file is fed into the network simulator whichthe importance of the message. In IEEE 802.11p MAC defines the realistic position of each vehicle during thelayer, there are six service channels and one control network simulation. The network simulator thenchannel and each of them has four different access implements the VANET protocols and produces a tracecategories. Consequently, during data transmission, there file which prepares complete information about the eventsare two contention procedures to access the medium: taking place in the scenario. Information is then analysed Internal contention procedure which occurs to evaluate the performance metrics of the IEEE 802.11p inside each channel between their access in VANET. categories by using the contention parameters VanetMobiSim is selected as a traffic simulator for this (Arbitrary InterFrame space (AIFS) and paper, since it is an open source and is validated against Contention Window (CW)). commercial simulators. This simulator supports Intelligent DriverModel with Intersection Management (IDMIM) which generates realistic vehicular mobility model .
Jiang et al.  mention that vehicular safety 170communications based on IEEE 802.11p consist of safety 160 150broadcast messages between neighbouring vehicles. 140 130Consequently, the overall IEEE 802.11p performance is 120 Distance (m) 110related to broadcast messages reception performance. PBC 100 90agent is a broadcast message generator implemented in ns- 80 702 version 2.34. We used this agent in order to define the 60 50broadcast message generation behaviour in our simulation. 40 30 20The scenario is a highway of 1500 metres long with three 10 0lanes in one direction and nine vehicles moving in these 0 6 12 18 24 30 36 42 48 54 60three lanes. The maximum speeds of the lanes are around Vehicle 280, 100 and 130 km/h respectively. The speed limit for Simulation Time (s) Vehicle 4each lane is 60 km/h. The distance between each lane is 4metres. In the scenario, an ambulance is in the emergency Vehicle 10situation travelling in the same direction as other vehicles Figure 3. Distance between the ambulance and other vehicles duringat the speed of 150 km/h. The ambulance is located behind movementother cars which are 100 metres apart. The IDMIMgenerates realistic vehicular mobility model. The Packet loss (%)ambulance transmits one periodic broadcast message witha payload of 250 bytes in every 0.2 seconds. In order toevaluate the effect of different message sizes on theperformance metrics, we implemented another twoscenarios in which the ambulance transmits periodbroadcast messages with the payload of 500, 1000 bytesrespectively. Each network simulations run twenty timeswith the same mobility trace to obtain an average and geta notion of statistical significance. ▬Packet loss between vehicle 1 and 2 ▬Packet loss between vehicle 1 and 4 Simulation Time (s) ▬Packet loss between vehicle 1 and 10 Figure 4. Packet loss between the ambulance and other vehicles during movement Figure 2. Scenario IV. RESULTS Fig. 5 demonstrates the throughput of vehicles 2, 4, and 10 with the message size of 250 bytes. The figure shows Results obtained from the scenario previously described that the throughput of vehicles 4 and 10 fluctuate betweenare presented in this section. Throughput, End-to-End 1.8 and 2.2 Kbps, when the distances between the vehiclesdelay, and packet loss were calculated for nine vehicles as and the ambulance are less than 138 metres. It can be seennumbered in Fig. 2 during the simulation run-time (i.e. 65 from Fig. 5 that all of the vehicles have nearly similarseconds). In addition, the impact of various speeds on throughput when the distances between vehicles anddifferent performance metrics was also evaluated. ambulance are less than 138 metres. In other words, Fig. 3 shows the distances between the ambulance and throughput of all the vehicles which their distances do notvehicles 2, 4, and 10 throughout the simulation . Also, exceed 138 metres from the ambulance are same and therePacket loss between the ambulance and these vehicles is no packet loss between these vehicles and ambulance.during the simulation time is illustrated in Fig. 4. It is The most important point is that each vehicle has differentclearly shown in Fig. 4 that there is no packet loss speed, as a result the throughput and packet loss are notbetween the ambulance and vehicle 4 after 58 seconds of affected by the varying speed.the simulation time; regarding to Fig. 3, the distance 6.4between the ambulance and vehicle 4 is less than 138 5.9metres after 58 seconds. Fig.4 shows that packet loss is 5.4dropped to 0% after 38 seconds of simulation, at the same 4.9 Throughput (kbps)time Fig. 3 demonstrates that the distance between 4.4 3.9ambulance and vehicle 10 is less than 138 metres after 38 3.4seconds. It provides similar results for vehicle 10 and 4. 2.9Accordingly, the vehicles can receive the broadcast 2.4 1.9message when their distance from the ambulance is less 1.4than 138 metres. 0.9 0.4 -0.1 0 0 10 20 30 40 50 60 Simualtion Time (s) Figure 5. Throughput of vehicle 2,4, and 10 (message size 250 bytes)
180 End-to-End delay between the ambulance and vehicles 1602, 4, and 10 with the message size of 250 bytes are shown Average Distance (m) 140in Fig. 6. A comparison between Fig. 3 and Fig. 6 shows 120that as long as the distance between vehicle and the 100ambulance is below 138 metres, the results of both figures 80look similar. As the distance between sender and receiver 60 40increases, End-to-End delay increases accordingly. It is 20observed that End-to-End delay is significantly influenced 0by the distance between sender and receiver of the 2 3 4 5 6 7 8 9 10message. As mentioned earlier, vehicles have different Vehicle Numbersspeed; consequently, various vehicle speeds do not haveany impact on End-to-End delay. Figure 8. Average distance between the ambulance and other vehicles (message size 250 bytes) 0.4665 100 0.46645 90 0.4664 80 End-to-End Delay (ms) Package Loss (%) 0.46635 70 60 Average 0.4663 50 0.46625 40 0.4662 30 0.46615 20 10 0.4661 0 0.46605 2 3 4 5 6 7 8 9 10 0.466 Vehicle Numbers 0 5 10 15 20 25 30 35 40 45 50 55 60 65 End-to-End Delay between vehicle 1 and 2 Simulation Time (s) Figure 9. Average packet loss between the ambulance and other End-to-End Delay between vehicle 1 and 4 vehicles (message size of 250 bytes) End-to-End Delay Between vehicle 1 and 10 Fig. 10 and Fig. 11 illustrate the average throughput and End-to-End delay with three different message sizes (250,Figure 6. End-to-End delay between the ambulance and other vehicles 500, 1000 bytes). According to these figures the average (message size 250 bytes) throughput and End-to-End delay are increased by increasing the message size, but the increment of Fig. 7, Fig. 8 and Fig. 9 illustrate the average throughput of vehicles 4, 7, and 10 is not as high as otherthroughput, distance and packet loss between all vehicles vehicles.and the ambulance respectively. The probability ofmessage reception for vehicles 4, 7 and 10 is less than 10other vehicles and they have the highest average packet Average Throughput (kbps) 9 8loss, since their average distance is more than other 7vehicles and at the beginning of simulation their distance 6 5from the ambulance is more than 138 metres. However 4other vehicles, which their distances do not exceed 138 3metres from the ambulance during simulation time, have 2 1equal and highest rate of average throughout without any 0packet loss. This is another reason indicating that 0 1 2 3 4 5 6 7 8 9 10throughput and packet loss are not influenced by different Message size 250 bytes Vehicle Numbersvehicle speed. Message size 500 bytes Message size 1000 bytes 1.8 Average Throughput (kbps) Figure 10. Average throughput of vehicles with different message sizes 1.5 1.2 0.9 0.6 0.3 0 2 3 4 5 6 7 8 9 10 Vehicle Numbers Figure 7. Average throughput of vehicles (message size 250 bytes)
 M. Amadeo, C. Campolo, and A. Molinaro, "Enhancing IEEE 1.6 802.11p/WAVE to provide infotainment applications in Average End-toEnd Delay (ms) 1.4 VANETs," Ad Hoc Networks, Elsevier, 2010. 1.2  WILLIAMS, B. “Intelligent Transport Systems Standards,” Artech 1 House Publishers, 2008. 0.8  S. Olariu and M. Weigle, Eds., “Vehicular Networks: From 0.6 Theory to Practice,” Chapman & Hall/CRC, 2009. 0.4  J. Härri, F. Filali, and C. Bonnet, “Mobility Models for Vehicular 0.2 Ad Hoc Networks: A Survey and Taxonomy,” research rep. RR- 0 06-168, Institut Eurecom, Mar. 2007. 0 1 2 3 4 5 6 7 8 9 10  D. Jiang, V. Taliwal, A. Meier, W. Holfelder, and R. Herrtwich, “Design of 5.9 GHz DSRC-based vehicular safety Message size 250 bytes Vehicle Numbers communication,” IEEE Wireless Communications, vol. 13, no. 5, Message size 500 bytes pp. 36–43, Oct. 2006. Message size 1000 bytesFigure 11. Average End-to-End delay between the ambulance and other vehicles with different message size V. CONCLUSION In this paper we studied the full details of the WAVEarchitecture and IEEE 802.11p standard for vehicular adhoc networks (VANET). We implemented the keyparameters of 802.11p in ns-2 network simulation usingrealistic vehicular mobility model generated byVanetMobisim traffic simulation. One scenario wasimplemented in the simulation. We analysed threeimportant metrics in order to evaluate the performance ofIEEE 802.11p standards. Based on our findings, we haveobserved that the performance metrics (throughput, End-to-End delay, and packet loss) are not affected by varyingvehicle speed. Analysis of throughput for the all vehiclesshowed that the probability of successful messagereception was same for all the vehicles when the distancebetween sender and receiver of the message was less than138 metres. In addition, End-to-end delay metric wasdirectly related to the distance between the vehicletransmitting the broadcast messages and its neighbouringvehicles. Results of scenarios with different message sizesdemonstrated that the average throughput and End-to-Enddelay metrics were increased by increasing message sizes. REFERENCES R. A. Uzcátegui and G. Acosta-Marum, “WAVE: A Tutorial,” IEEE Commun. Mag., May 2009. “Network Simulator ns-2,” http://www.isi.edu/nsnam/ns. “VanetMobiSim, ” http://vanet.eurecom.fr. S. Eichler, “Performance evaluation of the IEEE 802.11p WAVE communication standard,” in Proc. IEEE Vehicular Technology Conf., Baltimore, MD, US, Oct. 2007, pp. 2199-2203. T. Murray, M. Cojocari, H. Fu, “Measuring the performance of IEEE 802.11p using ns-2 simulator for vehicular networks,” in: Proc. IEEE EIT, 2008, pp. 98–503. K. Bilstrup, E. Uhlemann, E. G. Ström and U. Bilstrup, “Evaluation of the IEEE 802.11p MAC method for vehicle-to- vehicle communication,” Proc. IEEE Int. Symposium on Wireless Vehicular Communications, Calgary, Canada, Sept. 2008. S. Grafling, P. Mahonen, and J. Riihijarvi, “Performance evaluation of IEEE 1609 WAVE and IEEE 802.11p for vehicular communications,” in Proceedings of the 2nd International Conference on Ubiquitous and Future Networks (ICUFN ’10), June 2010, pp. 344 –348. Q. Chen, F. Schmidt-Eisenlohr, D. Jiang, M. Torrent-Moreno, L. Delgrossi, and H. Hartenstein, “Overhaul of IEEE 802.11 modeling and simulation in ns-2,” in Proc. 10th ACM Symp. MSWiM, Chania, Greece, Oct. 2007, pp. 159–168.