Effects of driving behaviour on vehicle cluster formation and its application

Node mobility in a vehicular network is strongly affected by driving behaviour, such as the choice of route. Although various route choice models have been studied in the transportation community, the effects of preferred routes on vehicular networks have not been discussed much in the networking literature. In this work, we set out to understand the effects of different parameters, such as route choice, traffic lights and the locations of road-side units, on the formation of vehicle clusters. We also show how the dynamics of vehicle clusters change over time using real-world vehicle trace data. Furthermore, inspired by the observation that the existence of vehicle clusters provides an opportunity to share Internet bandwidth among neighbouring cars, we propose an architecture that allows a car to utilise the unused Internet links of neighbouring nodes by stripping the data across multiple paths through its neighbours. We design a protocol, called enhancements for transmission control protocol (TCP) on a multi-homed environment (ETOM), to improve the performance of TCP in such an architecture without the need to modify the end points. We evaluate the performance of ETOM by comparing it with the concurrent multi-path transfer stream control transmission protocol (SCTP), and show that it can achieve similar or better performance.