Affinity Propagation-driven Multiple Weighted Clustering in MANETs

Abstract

This paper deals with a distributed clustering approach to tackle greedy clustering heuristics in MANETs. One of the most commonly used techniques to cluster nodes in a network is the Multiple Weighted Clustering (MWC) algorithm, which considers distinct heterogeneous performance metrics in a weighted form to select a Cluster-Head (CH). Since, clustering is a NP-hard problem, most of the MWC algorithms use greedy-clustering heuristics. The greedy approach intends to choose a strong, high priority node as a CH through frequent broadcast, by overlooking the topology evolution and the long term stability. The fundamental aim of this research is to address this gap and increase the efficiency of MWC technique in terms of stability, quality and cost of clustering. The MWC function used in this research considers node mobility factor, residual energy and connectivity. The Affinity Propagation (AP) method used in data mining is modified from a communication perspective and implemented to optimize the MWC function between nodes. The performance of the proposed approach is compared with the same MWC function with greedy approach using NS2. The simulation results show that the AP-driven MWC algorithm delivers better cluster stability and quality at a reduced clustering cost.