Packet Reachability of VANET in Bidirectional Road Scenario Panguo Fan Yuguang Li Guoqing Zhang Jinghua Li Dejun Mu Automation College of Northwesten Polytechnical University, Xi'an, China gniq@mail.nwpu.edu.cn Abstract-Bidirectional road is the basic and ordinary scenario of Vehicular Ad hoc Network (VANET) which is an emerging and challenging research applying mobile ad hoc technology to transportation system. According to diferent forwarding mechanisms in routing protocol of such network, the connections are divided into end-to-end (E2E) type and Store-Carry-Forward (SCF) type. The packet reachability probabilities, the representative of network connectivity, of both connection types are calculated and compared quantitatively using our proposed probability model. The results are validated by simulations in our experiments. Our work provides important and useful reference basis for network and routing protocol designers. Keywords- V ANET; network connectiviy; packet reachabiiy I. INTRODUCTION Vehicular Ad hoc Networks (VANET) represents a rapid emerging and challenging class of Mobile Ad Hoc Network (MANET). In such networks nodes (or vehicles) communicate with each other using equipped wireless transmission devices in hop by hop manner. This capability is particularly useul for distributing information such as trafic and safety information, weather information, and navigation information without the need of costly inrastructure. In order to facilitate the information dissemination in VANET, network connectivity is needed. The connectivity of VANET was studied by means of network simulation method in many literatures [1-4], while few of them analyzed connectivity theoretically. Connectivity of the network was discussed by percolation theory in [5]. Authors think the network almost sure cannot be ully connected for one dimensional dynamic network. The paper [6] considers VANET as a nominal system with disturbance. Under constant disturbance conditions, the lower bound of reachable neighbors for each vehicle to maintain a high connectivity is analytically derived only in one way street scenario. For bidirectional road scenarios in VANET, packets rom a source vehicle should be able to propagate and reach all the vehicles on the road segment, so packet reachability probability (PRP) is used to evaluate network connectivity in this paper. There are two different connection types in the existed routing protocols of VANET. The irst one is traditional end to-end connection always adopted by topology-based routing. It is called E2E connection in the following. In such case, packets sent from the source vehicle are forwarded by the intermediate node along the pre-calculated route to the destination. Once the forwarding node can not ind the proper 978-1-4244-6871-3/10/$26.00 ©201O IEEE 80 next hop for some reason, i.e, vehicle mobility, it has to drop the packet and the source node cannot communicate with the destination along the formed path any more. The second connection type is store-carry-forward (SCF) always used by opportunistic routing. Packets can be cached by the intermediate vehicle when failed to ind next hop node. When the new forwarding opportunity occurs, the packet is relayed thus it can be ransmitted to the destination at last. Our goal is to analyze the network connectivity of both connection types theoretically in terms of packet reachability probability. The validity of our theoretical analysis is also veriied by simulation. Conclusions are meaningul in designing and developing self organizing traic network. The reminder of this paper is organized as follows. In Section II we describe our probability model and assumptions. Section III analyzes packet reachability probability of E2E and SCF theoretically and Section IV is the experimental validation of our analysis. Finally, Section V concludes this paper. II. PROBABILITY MODEL AND ASSUMPTIONS The basic bidirectional road scenario is illustrated as Fig. 1. We propose a probability model to analyze the network connectivity probability in VANET. The connectivity probability is deined as the probability that packets can be transmitted rom the letmost vehicle to the rightmost vehicle in Fig. 1. In fact, many factors affect the packet transmissions of such inter-vehicle network including radio range, radio interference and time-varying vehicle mobility, etc. For simplicity, some assumptions are used in the model as follows: 1) The bidirectional road considered is long enough and its width can be ignored, so the network can be viewed as one dimensional structure. The vehicle length is also ignored. 2) Vehicle moving along the road can overtake each other like in multiple lanes. 3) All vehicles in the network has the same transmission range r and also two vehicles can communicate directly if the distance between them is less or equal to r. 4) Since vehicles are uniformly distributed on the road, for eficient large road length and the number of nodes, it can be shown that the amount of vehicles in any segment is considered as Possion distribution and the distance between two neighbors is exponential distributed [10]. The notations used in our model are deined as follows: 1) L: road length, unit is l. 2)t: vehicle density of the road in one direction, unit is #/m. 3) : total number of vehicles in the road and N = 2A L . 4) r: wireless radio range, unit is m.