Thesis (Ph.D.) - Cairo University - Faculty of Engineering - Department of Electronics and Communications

Using an efficient clustering algorithm is indispensable to solve the scalability and mitigate the effects of dynamic topology in vehicular ad hoc networks. Although current existing clustering algorithms show increased cluster stability, they still have some significant drawbacks to be tackled. In this thesis, we introduce a fully distributed mobility-based clustering algorithm.This algorithm increases clustering stability and efficiency by considering various metrics and schemes to form and maintain the clusters. In addition, to optimize the performance of the proposed algorithm in the vehicular environment, a many-objective optimization framework is used to automatically tune its configuration parameters.This framework is then extended to enable the proposed algorithm to dynamically tune its parameters in real-time. These parameters are adjusted based on the structure of the road and mobility characteristics of the vehicles. For effective performance of the algorithm in urban environments as well, we propose a machine learning-based approach to adapt the algorithm near intersections in the city. The proposed algorithm and approaches are simulated in realistic vehicular environments and have shown significant performance improvement in terms of stability and efficiency

Issued also as CD