The Research On Opportunistic Communication In Vehicular Ad Hoc Networks

Vehicular Ad Hoc Network(VANET) is the foundation of the Intelligent Transportation System, and will probably be an important part of the future “smart city”. Traditionally, VANET refers to the short distance wireless Ad Hoc networks formed between vehicles. However, in practical, due to the high mobility of vehicles and the complexity of radio propagation environment, it is difficult to maintain stable communication links between vehicle terminals. This makes its performance of communications capacity and message delivery distance is severely constrained. And the application fields are also limited. To overcome this challenge, the Opportunistic Networking technology can be used in VANET. It utilizes a message delivery mechanism which does not rely on the network topology information. This makes VANET more flexible and adaptable to meet the requirements of the data exchange in a variety of conditions.Opportunistic Networking uses the “Carry-Store-Forward” approach to delivery messages when the mobile nodes are close enough to each other’s wireless communication range. Compared to the traditional Ad Hoc network, it does not require the long existence of real-time end-to-end connections in the network, and thus meets the requirement of data delivery in the rapid-moving environment. For this reason, more and more researchers dedicate their attentions to this topic. Unfortunately, due to the principle of using delay as a trade-off for the network capacity, its delivery delay is high and the success of delivery always not guaranteed. In order to get a higher successful delivery rate and lower delivery delay, the replication-based multi-copy method is always used, at the cost of increasing the transmission overhead in the delivery process.In view of the characteristics such as fast-moving and high computing, we study on some key technologies of opportunistic networking in VANET to improve the delivery performance. We discuss the opportunistic routing problem from the following aspects: the uncertaintiness of opportunistic relay node in reducing delay, the characteristics of contact opportunities between vehicles, the characteristics of vehicles’ movement, as well as the impact of base station location on opportunistic delivery performance. The main contributions of this thesis include:Firstly, in view of the opportunistic characteristics that a relay node may reduce the forwarding delay or may not, we propose the concept of uncertaintiness to describe a relay’s effectiveness on reducing delay. Metrics Good Relay Ratio(GRR) and Delay Reduced Ratio(DRR) are defined to mesure this effectiveness. Based on the concept of uncertaintiness, we establish a probabilistic analysis model to investigate the benefit of multi-hop opportunistic forwarding. From the empirical study, we draw the following conclusions:(i) if direct forwarding delay is small, using relay node will not reduce the delivery delay;(ii) multi-copy schemes can greatly reduce the delivery delay. From the multi-hop benefit model, we obtain the expression of max-benefit count for three different forms of multi-hop forwarding scheme.Secondly, we propose the contact history-based opportunistic forwarding algorithm RED, which uses the Residual Expected Delay as relay selection metrics. This metrics can identify good relay nodes more accurately than the usually used Mean Inter-Contact Time. To further improve the accuracy of good relay selection, we devise a thresholdbased relay selection strategy in the RED algorithm. Evaluation results suggest that a reasonable threshold value can get more accurately results in finding out good and effective relay nodes. And it also demonstrates that RED based algorithms have better performance than the typical Spray&Wait and Max Prop algorithms.Thirdly, mining the movement coverage geographic characteristics of vehicle, we propose the MCCF algorithm. This algorithm utilizes the observation that vehicles have similar visit features as well as tighter contact opportunities in the same time. To exploit this characteristic, we establish a similarity measurement model and propose a feature extraction algorithm to compress the model overhead. Based on the similarity, MCCF algorithm combines the forwarding- and replication-based message delivery method, which not only can quickly spread message. but also limit the message diffusion range by dynamically adjusting algorithm parameters.Fourthly, we propose a base station location planning model and the LW algorithm based on the local weighting concept. The model and algorithm are configurable and flexible. According the input parameters and constraints, they can output either uniform or non-uniform base station location planning schemes. In the evaluation, we compare the performance of throwbox assisted message delivery that rely on three different types of hotspot, and draw the conclusion that using the hotspot of destination node can get better latency performance than the hotspot of source node and the common one.These research works are helpful for long-distance and large-range data delivery in VANET, and they are expected to provide technical support for a diverse applications and services in the future VANET.