MRV-M: A Cluster Stability in Highway VANET using Minimum Relative Velocity based on K-Medoids

Abstract

The cluster stability is the main feature to be achieved in the vehicular ad-hoc network. There are two primary algorithms in clustering to support cluster stability. These algorithms are K-Means and K-Medoids. They have a purpose of aiming cluster and selecting the Cluster Head (CH). The roadside unit (RSU) in the cluster has a responsibility to select the CH. In general, a central vehicle in a cluster will be selected as a CH. It is not always true whenever the CH has a different velocity with its members and it can influence the cluster stability. To handle this problem, we proposed a new method using relative velocity sorting that was inspired by K-Medoids algorithms. This proposed method is called as Minimum Relative Velocity based on K-Medoids (MRV-M). It combines the velocity and the position of each vehicle to select the best CH. The impact of M-RVM is increasing the stability of the CH duration. It also can increase the CH duration than using the previous method with the original algorithm from both of K-Means and K-Medoids. The proposed method was proved by the stable duration of the CH although it moves in different velocity with its members. Furthermore, it also can be proven in the small cluster size in each cluster. This has a consequence of network performance improvement, especially in throughput.