Efficient Clustering of Cognitive Radio Networks Using Affinity Propagation

Cognitive radios must be able to form collaborative wireless network clusters in dynamically changing environments to achieve such desired objectives as interference resilience and low communications overhead. In this work, we explore the affinity propagation (AP) message-passing technique to efficiently group nodes in an ad hoc cognitive radio network (CRN). With the proposed approach, nodes exchange local messages with their immediate neighbours until a high quality set of clusterheads and a corresponding cluster structure emerges. The messages are calculated based on measures of similarity between the network nodes, which are selected based on application requirements and the objective of the grouping process. As an initial application, we focus on finding a small dominating set of a CRN with the aim of reducing the number of nodes that participate in key network functions such as resource management and routing table maintenance. To demonstrate the merits of the proposed clustering approach, the AP technique is evaluated on randomly generated open spectrum access network scenarios. The simulation results demonstrate that the proposed technique provides a smaller number of clusters than methods based on approximating a minimum size dominating set of the corresponding ad hoc network graphs.