The 8th Annual IEEE Consumer Communications and Networking Conference - Special Session Information Dissemination in Vehicular Networks Operational Inferences on VANETs in 802.16e and 802.11p with Improved Performance by Congestion Alert *R. Bhakthavathsalam and Starakjeet Nayak *Supercomputer Education and Research Center, Indian Institute of Science, Bangalore-560012, India Tel: +91 80 22932940 Fax: +91 80 23602648 bhaktha@serc.iisc.ernet.in Birla Institute of Technology and Science, K.K. Birla Goa Campus, Goa-403726, India Tel: +91 9916505183 starakjeet.nayak@gmail.com Abstract Mobile WiMAX is a burgeoning network technology with diverse applications, one of them being used for VANETs. The performance metrics such as Mean Throughput and Packet Loss Ratio for the operations of VANETs adopting 802.16e are computed through simulation techniques. Next we evaluated the similar performance of VANETs employing 802.11p, also known as WAVE (Wireless Access in Vehicular Environment). The simulation model proposed is close to reality as we have generated mobility traces for both the cases using a traffic simulator (SUMO), and fed it into network simulator (NS2) based on their operations in a typical urban scenario for VANETs. In sequel, a VANET application called Street Congestion Alert is developed to assess the performances of these two technologies. For this application, TraCI is used for coupling SUMO and NS2 in a feedback loop to set up a realistic simulation scenario. Our inferences show that the Mobile WiMAX performs better than WAVE for larger network sizes. The long range is due to the transmission of signals at high power rates. Traffic prioritisation in Mobile WiMAX ensures high QoS, especially in voice and video applications. Deployment of Mobile WiMAX does not require infrastructure like building fibre or copper lines, hence the cost per bit of usage is propounded to be less if there are significant number of subscribers. WAVE, the IEEE 802.11p standard is similar to Mobile WiMAX in terms of mobility and data rates, but lacks proficient QoS support and has shorter transmission range. Figure 1 shows a graphical comparison of IEEE 802.11p and IEEE 802.16e in a normalized unit scale. In this paper, we propose Mobile WiMAX as a network opportunity for VANETs and analyse various performance metrics. Keywords VANET, Mobile WiMAX (IEEE 802.16e), WAVE (IEEE 802.11p), Packet Loss Ratio, Mean Throughput, Street Congestion Alert, TraCI I. INTRODUCTION VANET i.e. Vehicular Ad-Hoc Network has facilitated numerous safety and commercial applications for vehicular communication. Challenges for systems in vehicular environment are reasonably different than other wireless systems, such as high speed mobility, non- random movement, etc. In order to meet such challenges, a robust network technology is needed for VANET applications to ensure proper road safety and travel convenience. Currently WAVE (Wireless Access in Vehicular Environment IEEE 802.11p standard) and Mobile WiMAX (IEEE 802.16e standard) are the network technologies having capabilities suitable for VANETs. In this paper we make a comparative study of these two technologies and present their performance assessment using realistic simulations. Mobile WiMAX is a mobile extension of WiMAX, which aims at providing wireless broadband for metropolitan area networks. This technology is optimised for dynamic mobile radio channels and supports handoffs and roaming up to 100 kmph. Mobile WiMAX has long transmission range (~ 15kms in urban scenario) and provides data rates up to 35Mbps [1]. Figure 1. Comparison of Mobile WiMAX and WAVE As proposed in [2], realistic mobility models are the key components for VANET applications ensuring that the simulation results are carried through to real deployments. There are two such approaches for achieving realistic simulation: Network Centric and Application Centric [3]. In network centric model, a network simulator uses mobility traces generated by a traffic simulator for simulation, while in application centric model a feedback loop is established between traffic and network simulator, resulting in a dynamic real time simulation. We have used both the simulation approaches in our analysis. 978-1-4244-8790-5/11/$26.00 2011 IEEE 467
The rest of the paper is organised as follows: Section 2 contains the related work on Mobile WiMAX, WAVE and TraCI. In Section 3, we describe the simulation model and present the results based on the experiment. Section 4 explains the VANET application Street Congestion Alert. Finally in Section 5 we conclude our results with a view on the future trends. II. RELATED WORK A. Mobile WiMAX - IEEE 802.16e Mobile WiMAX is one of the wireless broadband standards capable of providing the Quadruple play Technologies Data, Voice, Video and Mobility using a single network [4]. In other words, this can support quality broadband service in a mobile set-up, making it one of the strongest contenders for VANETs. Modifications in its physical (PHY) layer and medium access control (MAC) layer make Mobile WiMAX suitable for such applications. In PHY layer, it supports OFDM (Orthogonal Frequency Division Multiplexing) and OFDMA (Orthogonal Frequency Division Multiple Access), which increase the bandwidth by splitting broad channels into narrow channels each narrowband carrying different frequency that can carry different part of a message separately [5]. In the MAC layer, Mobile WiMAX supports two modes point to multipoint (PMP) and mesh. PMP organises nodes into cellular-like structure consisting of base stations (BSs) and subscriber stations (SSs) [6]. The channel is split into uplink (SS BS) and downlink (BS SS). On the other hand, in mesh mode mobile nodes can act as relaying routers - in addition to their senders and receivers role - forming an ad-hoc network between the nodes. In this paper we have used OFDM support for PHY layer and PMP mode for MAC layer for our simulations. Handoff, i.e. switching from one base station to another is of prime importance in Mobile WiMAX. Handoffs are broadly classified as soft handoff (make-before-break) and hard handoff (break-before-make). Several mechanisms like FBSS (Fast Base Station Switching), MDHO (Macro Diversity Handover) have been proposed [7]. Hard handoffs can be very efficient is terms of channel usage, as only one channel is occupied at a time. However this can cause more delay and can be problematic for traffic like video streaming. On the other hand, soft handoff is reliable as the link is never broken, but it can cause reduced throughput as it has to support multiple channels sharing the data stream simultaneously. There are few challenges faced by Mobile WiMAX, one of them being spectrum allocation as there is no global licensed spectrum for WiMAX. Currently Mobile WiMAX operates at 2.3 GHz in the Asia-Pacific region, 2.5 GHz in the United States and 5.5 GHz in the European Union. Another roadblock in the full deployment of WiMAX is that lack of Internet-backbone bandwidth in order to provide backhaul necessary for supporting large number of subscribers. B. WAVE IEEE 802.11p IEEE 802.11p, also known as WAVE is a draft amendment to the IEEE 802.11 standard, incorporating applications to fast changing vehicular networks. This standard is used as groundwork for Dedicated Short Range Communication (DSRC) [8]. It operates in the 5.9 GHz band and supports both vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications. The maximum data rate supported by this standard is 27 Mbps as against 54 Mbps of other 802.11 flavours. The mobility support is up to 100kmph, making this suitable for VANETs applications involving highway scenario [9]. In this document we compare WAVE with Mobile WiMAX on a similar simulation background. C. TraCI Towards realistic simulation TraCI Traffic Control Interface is an interface enabling application centric simulation. It develops a feedback loop mechanism between traffic simulator (SUMO) and network simulator (NS2) [10]. Using this mechanism, the traffic simulator feeds the network simulator with position updates of vehicles and the latter issues mobility commands for vehicles to the traffic simulator (for e.g. ChangeRoute, Stop, Start etc.) Hence the two simulators work in conjunction to provide a realistic simulation scenario, ideal for testing VANETs applications. III. SIMULATION MODEL AND RESULTS A. Simulation Set-up We define the goal of simulation as performance evaluation of IEEE 802.11p and IEEE 802.16e in a typical urban scenario. The simulation environment consists of a network of two-lane roads meeting at a junction as shown in Figure 2. We set up Vehicle to Infrastructure communication by placing four base stations for each of the road segments. In order to get a realistic simulation model, SUMO is used to generate mobility traces of vehicles. These traces are fed into NS2 for simulation. Roads are populated with vehicles incrementally, starting from 20 nodes up to 100 nodes. Figure 2. The simulation scenario for performance evaluation of VANETs 468
B. Simulation Parameters The simulation is run for two cases, first by using Mobile WiMAX nodes, and second by using WAVE nodes. Vehicular mobility model parameters for SUMO are listed in Table 1. The network simulator parameters are described in Table 2 for IEEE 802.11p [11] and IEEE 802.16e. We have used 64 ¾ QAM as the modulation scheme for both the networks as this produces maximum throughput. The mobility module for 802.16e is downloaded from NIST [12] and 802.11p patch is obtained from [13]. As seen from the plot, throughput per node decreases for both the MAC types as the number of vehicles increase in the network. Initially mean throughput for WAVE is observed to be more than WiMAX for lesser number of vehicles. As the network size expands more (for e.g. >40 nodes in this case), the mean throughput of WAVE nodes drop more rapidly than compared to WiMAX nodes. Thus WiMAX is found to be more stable than WAVE for larger vehicular network in terms of mean throughput. TABLE I VEHICULAR MOBILITY MODEL PARAMETERS Parameters Length of Route 5020m Number of road segments 4 Max. speed of vehicles 15.56 m/s Total no. of vehicles 20 to 100 TABLE II NETWORK SIMULATOR PARAMETERS Parameters (802.11p) (802.16e) Radio propagation model Propagation/ Nakagami Propagation/ TwoRayGround Network interface type Phy/ WirelessPhyExt Phy/ WirelessPhy Modulation used 64 ¾ QAM 64 ¾ QAM Routing protocol AODV AODV Packet Size 1500 Bytes 1500 Bytes Traffic Type UDP/CBR UDP/CBR Simulation time 350s 350s Figure 3. Graph of Throughput per node vs. number of vehicles for IEEE 802.11p and IEEE 802.16e 2) Packet Loss Ratio: Packet Loss Ratio is defined as: Packets dropped due to collision Packets dropped + Packets received Packet loss due to collision is an inevitable phenomenon. Here we compute packet loss ratio in order to determine the efficacy of the network type. The plot between packet loss ratio and number of vehicles is shown in Figure 4. The simulation is run for 350 s, and the number of vehicles in the network is continuously varied for both the cases. Vehicle distribution in the scenario is not random, but is specified by SUMO, so as to have uniformity with respect to the base stations. C. Performance Criteria and Results Mean Throughput per node and Packet Loss Ratio are the performance metrics we intend to measure in this experiment. 1) Mean Throughput: Mean Throughput is essential for data intensive VANET applications like video streaming. High throughput is desired for quality data transfer. In this scenario, we compute mean throughput per node against number of vehicles for Mobile WiMAX and WAVE. The plot is shown in Figure 3. Figure 4. Graph of Packet Loss Ratio vs. number of vehicles, measured in the simulation set-up 469
From Figure 4, it is evident that packet drop due to collision is more severe for IEEE 802.11p than IEEE 802.16e as number of vehicles increase. This loss of packets due to collision contributes to lesser mean throughput for larger vehicular network in case of WAVE as compared to WiMAX. This result compliments the mean throughput curves in the previous section. IV. VANET APPLICATION STREET CONGESTION ALERT In this section we describe a VANET application called Street Congestion Alert, which is developed to compare the performances of Mobile WiMAX and WAVE on VANETs. In a typical urban scenario, slow traffic movement due to congestion in a street is common. In many cases taking an alternate route to reach destination takes lesser time than moving through the congested street. Street Congestion Alert application alerts the vehicles regarding congested streets, hence helping the drivers to take alternate routes to reduce the travel time. For the application, we have taken a portion of Manhattan city roads as depicted in Figure 5. The streets marked in purple are congested streets and the ones marked in red depict the fast moving alternate route to the destination. A Base Station has been placed at the junction that broadcasts the details regarding the congested streets. As vehicles move towards the junction, they receive the broadcast packets and take the alternate route to save time. Few vehicles that do not receive the message due to network collision take the slow traffic street instead of the alternate route. At the end we measure the total time taken by the vehicles to reach the destination. The application is run for two cases, first by using IEEE 802.16e nodes and second by using IEEE 802.11p nodes. mechanism, the routes of vehicles can be dynamically altered when they receive the message from base station. Other simulation parameters are listed in Table 3. A simulation snapshot of the application is given in Figure 6. TABLE III MOBILITY PARAMETERS FOR STREET CONGESTION ALERT Parameters Traffic Simulator SUMO 0.9.8 Number of road segments 6 Max. speed of vehicles (normal streets) 15.56 m/s Max. speed of vehicles (congested streets) 4.21 m/s Total no. of vehicles 20 to 100 Broadcast packet size 100 Bytes Broadcast Interval 0.25 s Figure 6. Simulation snapshot of the application Figure 5. Map showing source and destination, congested short route and the alternate route. For simulation, roads are populated by varying the number of vehicles, starting from 20 up to 100. We have used TraCI, which creates a feedback loop mechanism between traffic simulator (SUMO) and network simulator (NS2). With this Figure 7. Graph showing time taken by vehicles to reach the destination for IEEE 802.11p and IEEE 802.16e nodes 470
As we can see from Figure 7, time taken by IEEE 802.11p nodes increase more as compare to IEEE 802.16e nodes as the number of vehicles increase. This can be attributed to the fact that as the size of vehicular network increases; the efficiency of data transfer of former reduces as compared with its latter counterpart. Hence more vehicles implementing IEEE 802.11p fail to receive the broadcast message and take the slow route, delaying the overall time to reach the destination. V. CONCLUSION To begin with we have portrayed the operations of VANETs using Mobile WiMAX for real life vehicular applications. Subsequently the similar performance of VANETs is assessed adopting 802.11p in WAVE. The Mean Throughput and the Packet Loss Ratio are the two metrics evaluated for both the cases through simulation techniques. The measured values of both the metrics substantiate that Mobile WiMAX is a better networking technology for VANETs with large network size, which is typically the case. Further to demonstrate this inference, we have developed a VANET application called Street Congestion Alert. The idea was to measure the total travel time of vehicles when street congestion is invoked, given a stationary base station broadcasting the information about congestion. The simulation renders that more number of WAVE nodes fails to receive broadcast message as compared to that of Mobile WiMAX and hence do not alter their routes to save time by avoiding delay due to congestion. Thereby the total travel time is found to be more for WAVE than Mobile WiMAX, which is in corroboration with our previous results. Therefore Mobile WiMAX is preferred as a better network opportunity for VANETs, as also bearing in mind that the cost per bit of usage is low for 802.16e as compared to other wireless technologies. ACKNOWLEDGMENT The authors sincerely thank the authorities of Supercomputer Education and Research Center, Indian Institute of Science, Bangalore for the encouragement and support. REFERENCES [1] http://www.unwired.ee.ucla.edu/dsrc/dsrc_testbed_simple.htm [2] R. Bhakthavathsalam, Starakjeet Nayak, Srikumar Murthy G, Expediency of Penetration Ratio and Evaluation of Mean Throughput for Safety and Commercial Applications in VANETs, Proceedings of Nets4Cars 09, October 2009. [3] M. Pi orkowski, M. Raya, A. Lezama Lugo, P. Papadimitratos, M. Grossglauser and J.-P. Hubaux, TraNS: Realistic Joint Traffic and Network Simulator for VANETs, ACM SIGMOBILE Mobile Computing and Communications Review, Volume 12, Issue 1, January 2008. [4] Venkat Annadata, 802.16e & 3GPP Systems Network Handover Interworking, Tech Mahindra Limited, April 2010. [5] Steven J. Vaughan-Nichols, Mobile WiMax: the Next Wireless Battleground Computer, vol. 41, no. 6, pp. 16-18, June 2008. [6] Kejie Lu, Yi Qian, Hsiao-Hwa Chen, A Secure and Service-Oriented Network Control Framework for WiMAX Networks, IEEE Communications Magazine, May 2007. [7] Hoang N, Kawai M, An Adaptive Hanoff Scheme for Mobile Wimax Networks, Wireless VITAE 2009. [8] http://en.wikipedia.org/wiki/ieee_802.11p [9] Stephan Eichler, Performance Evaluation of the IEEE 902.11p WAVE Communication Standard, WIVEC 2007. [10] http://sumo.sf.net/wiki/index.php/traci [11] Jiang D. and Delgrossi L., IEEE 802.11p: Towards an International Standard for Wireless Access in Vehicular Environments, Vehicular Technology Conference, pp. 2036 2040, VTC Springer, 2008. [12] http://w3.antd.nist.gov/seamlessandsecure/pubtool.shtml [13] http://dsn.tm.uni-karlsruhe.de/english/overhaul_ns-2.php 471