On the throughput performance of cluster-based cognitive radio networks

This paper addresses effect of reporting channel bandwidth on cognitive radio (CR) networks. A cluster based approach is considered where the secondary base station is replaced by a fusion center and a global reporting channel is used instead of local ones. A new approach to select the fusion center based on the general centre scheme in graph theory is proposed. The minimal dominating set (MDS) clustering approach is used to minimise the set of clusters that keeps the network connected. The effect of various parameters such as cluster size and number, quality of the reporting channel and sensing time on sensing efficiency, accuracy and per node throughput are investigated. Results show cluster based cooperative sensing throughput outperforms conventional cooperative sensing especially when the reporting channel has high probability of error. Systematic ways to determine optimum number of clusters and optimum sensing time are developed.