Abstract
Vehicular Ad-Hoc Networks (VANETs) is an emerging technology which is being used in wide areas of applications especially to provide safety and comfort to the users sitting in the vehicles. Vehicles in VANETs act as intelligent machines for taking adaptive decisions for information dissemination from source to destinations. Vehicles can send warning messages to drivers sitting in other vehicles about varying traffic conditions to increase safety and user convenience on roads. But due to constrained mobility patterns as well as peculiar traffic conditions of vehicles, the timely delivery of messages to their final destination is a challenging task in VANETs. Clustering is one of the most effective ways of managing and stabilizing such networks. In this paper, a learning automata based clustering scheme is proposed for vehicles in VANETs. The concept of number of links and vehicular mobility is used for cluster formation and cluster-head selection. During the cluster formation process, nodes with relatively higher degree of connectivity initially form a backbone which is designated as leadership. The leadership nodes then participate in cluster-head election and efficient cluster reorganization using aggregate degree of connectivity of vehicles. Simulation results depict the effectiveness of proposed algorithm on account of its comparable cluster stability with existing protocols especially in urban scenarios.