Clustering in VANETs under realistic channel conditions

A Vehicular Ad hoc Network (VANET) differs from conventional wireless networks by its lack of infrastructure. Cars equipped with wireless devices must harmonise their transmissions in order to achieve the objectives of the network. Clustering serves as a means of doing so, by grouping vehicles according to spatial distribution and relative velocity. These groups, led by an elected cluster head, can serve as the foundation of routing, collision detection, and infotainment applications. After more than twenty years of research, clustering has diversified into a wide array of approaches and mechanisms, all of which have the potential for wide-spread application. However, the research has lacked crucial elements. Firstly, published taxonomies tended to compare algorithms and protocols based on their application. There is minimal emphasis on the underlying design choices and their suitability to the problem of VANET communication. Secondly, channel modelling practices have been insufficient to adequately validate new designs. Without these two elements, new avenues for research have gone unnoticed. This thesis aims to address these issues.