A PROBABILITY COLLECTIVES APPROACH TO WEIGHTED CLUSTERING ALGORITHMS FOR AD HOC NETWORKS

摘要

The Weighted Clustering Algorithm has been proposed for mobile ad hoc wireless networks to select a subset of the nodes to be local purveyors of media access control services to the rest of the nodes. The nodes selected are called clusterheads, and each clusterhead services the other nodes within its radio range. The underlying optimal assignment problem is NP-hard. We introduce a new metaheuristic known as probability collectives to the problem of assigning clusterheads. This is an agent-based approach, with each node in the network acting in a quasi-autonomous way. This new approach shows improved clustering results, particularly with regard to the distribution and balance of cluster sizes.