Lowest weight: Reactive clustering algorithm for adhoc networks

Abstract

In this paper, we address clustering in ad hoc networks. Ad hoc networks are a wireless networking paradigm in which mobile hosts rely on each other to keep the network connected without the help of any pre-existing infrastructure or central administrator. Thus, additional features pertinent to this type of networks appeared. In fact, centralized solutions are generally inadaptable due to the need for cooperative network operations. To ensure efficient, tolerant and durable cooperative operations, nodes need to organize themselves. Clustering is an organization method which consists in grouping the nodes into clusters (groups) managed by nodes called clusterheads. In this paper, we present existing clustering algorithms and propose a new solution inspired from two of these algorithms (Lowest Id and WCA). This solution, called Lowest Weight, exploits their advantages and relieve to their drawbacks in terms of clusters stability and computational overhead. Simulation experiments were conducted to evaluate the performance of the algorithm proposed in terms of clusters numbers, clusterheads lifetime and the number of reaffiliations (node moving from a cluster to another). Results show that Lowest Weight ameliorate performs of existing algorithms especially regarding mobility leading to more suitable, adaptable, scalable and autonomous clustering. © IFIP International Federation for Information Processing 2006.