Performance Evaluation of VANETs with Multiple Car Crashes in Different Traffic Conditions Georgios Charalampopoulos 1,2 and Tasos Dagiuklas 1 1. Dept. of Computer Science, Hellenic Open University, Greece, Emai:{gcharalampopoulos, dagiuklas}@eap.gr 2. Dept. of ECE, Universty of Patras, Greece, Email: gcharalampopoulos@eap.gr Abstract. This paper studies the performance of VANETs with multiple car crashes under different traffic conditions. More specifically, a module has been designed and developed in a simulation environment in order to generate multiple vehicle crashes. It is shown that the VANET network statistics are affected by the node density and the number of vehicle crashes. Keywords: VANET, Car Accident, VANET Simulation Tools 1 Introduction There is a growing interest in deployment of warning services in Vehicular Ad-Hoc Networks (VANETs) for urban contexts allowing the communication among vehicles [1], [2]. These networks can be used for a variety of applications such as traffic management on motorways, emergency services and the prevention of collisions among vehicles. Research activities so far are mainly focused on approaches to reduce traffic congestion and provide general information services [3]. Several mobility models in VANET environment have been studied and simulated to extract useful information regarding different routing protocols, beaconing and delivery of messages etc. The main goal of this paper is to design a module in a simulated environment allowing the creation of crashes/collisions among the vehicles. Important parameters of this car-crash module are the following: Node density, Departure Time - Arrival Time of each node.[4] The aim is to investigate the impact of the car crashes on the mobility models and the VANET network metrics. Moreover, we study the impact of the number vehicles involved in the collision on the performance of VANET. The paper structure is the following: VANET Mobility Modeling, Simulation Scenarios, Results and Conclusion. 2 VANET Mobility Modeling For the VANET simulation scenarios, realistic mobility models in microscopic simulation (it takes into account road and node characteristics) have been used providing P29-1
road maps, the number of travelling cars and some road and car parameters such as maximum car speed, road limitations, departure and arrival times of each car [5]. Amongst others, the most common simulator environments for these mobility models are SUMO - Simulation of Urban MObility, VanetMobiSim and STRAW [6]. Without loss of generality, SUMO model has been used. Node density is one of the factors that affect the mobility models and the vehicular networks used for the simulation. The different node densities are used to represent the different traffic conditions for peak and off peak hours. A parser has been created to calculate the road network that will be used in accordance with the street map to carry out the simulations. Fig. 1. Process of Road Network Calculator Parser The figure above shows the process of calculating the road network using the XML map file that has been generated by SUMO. The road network is calculated from the sum of the distances of all the road-ids and the distance of the road network. Then, using node density [7], and the road network, the process calculates the node density for the simulation scenarios that will be presented below. Vehicle crashes are another important factor that has been influenced by the mobility model and the node density. When there are accidents in VANET, the vehicles around the point of interest (crash event) are mainly influenced by stopping and/or changing direction to reach the final destination using the Dynamic Re-route provided by SUMO. [6] Furthermore, the network infrastructure is influenced by the number of nodes, because emergency messages from crashed vehicles are sent after the accident in order to inform drivers and emergency response teams such as police vehicles, ambulances etc when required. To find vehicles in a simulated environment that can crash among each other, another parser written in JAVA has been created. This parser initially selects a node randomly and then searches for other nodes that have common edges on their fixed routes with the first node and are close with it in distance. As an effect, a vehicle crash is created. The picture below shows the process of finding crashes of vehicles. More specifically, the parser uses XML route file, which is loaded for use in the simulation. Then the parser searches for the previously mentioned nodes in order to create the accident. Otherwise the first random selected node is referred as a broken-down vehicle. For P29-2
multiple car crashes, the JAVA parser will be compiled several times to extract more than one crash. Fig. 2. Process of Car Crashes Parser 3 Simulation Scenarios For Vehicle Crash simulation, Veins framework [8] has been selected because it integrates the OMNeT++ and SUMO [9] and can offer online re-configuration and rerouting of cars in reaction to the network simulator, IEEE 802.11p and IEEE 1609.4 DSRC/WAVE network layers, supporting realistic maps and traffics. The IEEE 1609.4 defines the multi-channel and QoS operation of radios; vehicles with a single radio will periodically switch among multiple channels. [10] Firstly, a realistic map and the road network are imported from OpenStreetMap.com [11], the routes of the cars are generated using SUMO and then exported to the network simulator. OMNeT++ considers all the cars as nodes and simulates the scenario. If there are changes in the network, Veins can modify the scenario in SUMO and the realistic traffic of the city center is simulated as a VANET in OMNeT++. Simulation scenario is done using the configuration, dataset and parameters for 2000 seconds. About 667 cars for peak and 383 cars for off peak hours travel in the map at this time P29-3
and each scenario runs the simulation 20 times starting from zero. Three different simulation scenarios have been created with one, two and three car crashes. Each scenario has been replicated 20 times and for each scenario a different start time has been used for each node. The following table shows network parameters for our simulations. Beaconing Rate 10 Hz CWmin 15 Max Transmission Power 20 mw CWmax 1023 Thermal Noise -110dBm CCH interval 50 Header Length 256 bit Data Rate 18Mbps Slot Duration 16 s Dedicated Frequency Range 5.9 GHz Table 1. Network Simulator Parameters 4 Results The results from the simulations scenarios will be presented below. Especially, Received Broadcast and Total Lost packets, Busy Time and Total CO2 Emission will be discussed to check how these metrics are affected by the different number of crashes. Fig. 3. Total Received Broadcast Packets The first graph shows the Received Broadcast Packets and nodes and that the different number of vehicles influences the network infrastructure. The value of Received Broadcast Packets is higher for peak than the off peak hours. Moreover, the number of crashes increases the number of Received Broadcast packets, which is due to the fact that more and more nodes close to the area of interest must be updated after the crash event. The graph below shows that different number of vehicles impact the number of Total Lost packets because vehicles collect more information when more than one crash has been occurred. For the peak hours the number of these packets is larger than that to the off peak hours. P29-4
Fig. 4. Total Lost Packets Busy Time is affected by the different number of vehicles and car crashes. When the number of car crashes increases, the Busy Time increases too. Fig. 5. Total Busy Time in seconds The final graph shows that the Total CO2 Emission per vehicle. As expected, this parameter increases when the number of accidents and node density increase because many vehicles stop and wait during their route. Fig. 6. Total CO2 Em. / Vehicle in mgr The results show that the multiple car crashes and the different number of vehicles in the same area are highly affecting the mobility model and the performance of the network infrastructure. P29-5
5 Conclusion This paper has studied how multiple car crashes can be generated in a VANET simulation environment. We have investigated the impact of node density and number of car crashes on VANET performance evaluation. Realistic scenarios in an urban area with real vehicle routes have been studied. The future work includes simulations of multiple car crashes using Vehicle-to-Infrastructure (V21) communication and comparative study with V2V communication and implementation of other applications regarding the warning services of VANETs. Acknowledgement. This work was supported by the SALUS project (grant number: 313296), funded by the EC FP7 ICT-security collaborative research 6 References 1. Sebastian Grafling, Petri Mahonen, and Janne Riihijarvi. Performance Evaluation of IEEE 1609 WAVE and IEEE 802.11p for Vehicular Communications. In International Conference on Ubiquitous and Future Networks, pages 344-348, 2010. 2. H. Conceicao, L. Damas, M. Ferreira, and J. Barros, Large-scale simulation of V2V environments, in Proceedings of the 2008 ACM symposium on Applied computing. ACM, 2008, pp. 2833. 3. C. Manasseh and R. Sengupta. Middleware to enhance mobile communications for road safety and traffic mobility applications. IET Intelligent Transport Systems, 4(1):24, 2010. 4. Harri, J.; Filali, F.; Bonnet, C., Mobility models for vehicular ad hoc networks: a survey and taxonomy, Communications Surveys & Tutorials, IEEE, vol.11, no.4, pp.19-41, Fourth Quarter 2009. 5. Karnadi, F.K.; Zhi Hai Mo; Kun-chan Lan;, Rapid Generation of Realistic Mobility Models for VANET, Wireless Communications and Networking Conference, 2007.WCNC 2007. IEEE, vol., no., pp.2506-2511, 11-15 March 2007. 6. SUMO, Simulation of Urban MObility http://sumo.sourceforge.net/ 7. Clean Air Initiative, http://cleanairinitiative.org/portal/greece 8. Veins, The open source vehicular network simulation framework, http://veins.car2x.org 9. A. Kopke, M. Swigulski, K. Wessel, D. Willkomm, PT Haneveld, TEV Parker, OWVisser, HS Lichte, and S. Valentin. Simulating wireless and mobile networks in OMNeT++ the MiXiM vision. In Proceedings of the 1st inter-national conference on Simulation tools and techniques for communications, networks and systems & workshops, pages 1-8. ICST, 2008. 10. IEEE Draft Standard for Wireless Access in Vehicular Environments (WAVE) - Multi-Channel Operation, IEEE Pl609.41D9, August 2010, pp. 1-61, 2010. 11. OpenStreetMap, http://www.openstreetmap.org P29-6