A Study on Cluster Stabilization Strategies for Human-Driven MANETs

Abstract

Clustering is a commonly used method to adequately manage the network resources provided by mobile ad hoc networks (MANETs). In order to work appropriately in these networks, clustering protocols integrate stabilizing strategies that cope with topological changes caused by node motion. Although there is a number of studies related to clustering, cluster stability remains as a critical feature that has been barely addressed. It is not known, for instance, how stability of clustering protocols is affected when network nodes move following human mobility patterns (human-driven MANETs). The aim of this paper is to explore this open issue in order to improve the performance of clustering protocols in a human-driven environment and, in this way, to advance MANET research one step toward the deployment of real networks. To this end, the performance of several stabilizing strategies is assessed considering a network scenario where nodes roam according to the self-similar least-action walk (SLAW) model, which integrates several statistical features of human motion. Additionally, this work proposes to classify the stabilizing strategies into clusterhead-election based strategies, strategies based on relaxed-validity policies and strategies controlling the number of clusters. Our findings show that clustering protocols have to integrate flexible policies in order to improve stability as well as strategies dealing with burst reaffiliations and big-sized clusters.