Graph cut based clustering for cognitive radio ad hoc networks without common control channels

Abstract

Clustering is an efficient tool to improve the routing and data transmission performance in large scale networks. However, in cognitive radio ad hoc networks (CRAHNs), clustering design is challenging due to the dynamic spectrum access and the blind information environment. In this paper, we propose a novel distributed clustering algorithm for CRAHNs, where neither a dedicated common control channel (CCC) nor prior topology information is required. First, a neighbor discovery protocol without relying on CCC is proposed to construct the local topology. Then, we model the network as a undirected graph and formulate the clustering process as a graph cut problem. We design a mincut based heuristic algorithm to approximate the optimal clustering solution. After this, we also present a synchronize protocol to achieve the global consistency of cluster memberships. Finally, we propose a proactive cluster maintenance mechanism to reduce the interferences caused by PU activities. We validate our work through comparisons with other clustering methods. The simulation results show that, by adjusting the cluster structure according to the changing spectrum, the proposed method reduces the interference and improves the network efficiency.