A Cluster-on-Demand Algorithm with Load Balancing for VANET

Cluster–on–Demand VANET clustering algorithm (CDVC) along with load balancing for urban is proposed. Urban vehicles are characterized by unpredictable moving direction. These challenges are solved by CDVC, which is composed of three main procedures; that is, initial clustering, cluster merging, and cluster head selection. In initial clustering, vehicles are clustered determines the boundary of each cluster. In cluster merging, Self–Organizing Maps (SOMs) is used for re–clustering by the similarity of nodes, which guarantees the stability of clusters. It leads to achieve load balancing. In cluster head selection, the information of location and mobility are combined to select a more stable cluster head. The performance of CDVC is evaluated and compared with the Lowest ID (LID) and Mobility based on clustering (MOBIC). Finally, the simulation results reveal that CDVC is superior to LID and MOBIC in terms of cluster head duration, clusters number, and load balancing.