Degree Name

Doctor of Philosophy


School of Electrical, Computer and Telecommunications Engineering


Vehicular Ad Hoc Networks (VANETs) are distributed, infrastructure-independent wireless networks that are formed between both stationary and moving vehicles. Despite over a decade of research, there exist few practical deployments of largescale vehicle to vehicle networks. The area of routing protocols is of particular interest since it forms the basis of inter-node communications in a distributed infrastructure-less network. Simulations are an important means of evaluating new VANET protocols and architectures since physical experiments require a large investment in resources and are beyond the reach of most researchers. Simulations rely on approximations of reality; the better the approximation, the more confidence may be placed in the results. The physical propagation model is one of the fundamental aspects of any network simulation since it defines how signals travel between nodes. A review of urban routing protocols revealed that minimal consideration has previously been given to the selection of a realistic propagation model when evaluating routing protocol performance. The use of simplistic propagation models was found to lead toa very inaccurate representation of performance, especially the relative performance difference between protocols. This thesis presents a series of experiments intended to characterise the propagation model in urban environments. The CORNER propagation model is studied and enhanced based on the data obtained from the experiments. A new greedy routing approach is then introduced, which uses knowledge of the signal propagation environment to make better routing decisions. Analytical results demonstrate the improved route-finding capabilities of the new approach over existing greedy and source-based routing protocols, with the new approach resulting in the selection of shorter paths in almost every situation. Simulation results establish an improvement of between 90% and 300% over GPSR, a well known stateless greedy routing protocol. A semi-stateful fall-back mechanism is then proposed to augment the greedy routing approach when it fails and is found to result in a further 10-15% improvement in packet delivery ratios over the purely greedy approach.