Study on vehicular network application and simulation

Abstract

<p>VANET is an emerging mobile ad hoc network paradigm that facilitates vehicle-to-vehicle and vehicle-to-infrastructure communication. The most important application of the VANET is for driving safety. Road condition-awareness is critical for driving safety. Existing VANET-based systems usually assume drivers detect and report safety related road conditions, which however may be untrue because, drivers may not be willing to perform these duties, or such duties may distract drivers and thus make driving even unsafe. Therefore, automatic detection without human intervention is desired. As the first contribution of this thesis work, an automatic road condition detection system has been designed based on the idea of collecting and analysing the footprints of vehicles to infer anomaly. It has also been studied how to utilize inexpensive roadside devices, such as sensors, to facilitate the information collection and analysis, especially in the absence of connectivity between vehicles.</p> <p>Due to the difficulty of conducting large-scale experiments on real roads, simulation plays an important role in VANET research. To make simulation close to the reality, it is desired to include detailed and realistic simulation of vehicle behaviour under various road conditions, and this is especially needed for studies targeted at driving safety. In the past, however, the simulation of vehicle behaviours are often overly simplified and implemented as a trivial extension of the network simulator. As a second contribution of this thesis work, a detailed and realistic simulator of vehicle behaviour has been developed based on the car-following and lane-changing models.</p> <p>As the simulation of vehicle behaviour and that of communication behaviour are different tasks, they should be implemented separately for better modularity and meanwhile they should be seamlessly integrable. As another contribution of this thesis work, the online and seamless integration of vehicle behaviour simulator and network simulator has been studied. Specifically, a set of APIs has been designed and implemented atop the vehicular behaviour simulator to facilitate its integration with network simulator. Being a concrete example, the integration of ns2 and SUMO, an open-source vehicular behaviour simulator, has been implemented, and applied to simulate an electric vehicular network.</p>