Cooperative spectrum sensing over correlated log-normal channels in cognitive radio networks based on clustering

In this paper, the problem of cooperative spectrum sensing in cognitive radio networks based on linear combination of local observations is considered. In particular, log-normal shadow-fading is considered in both sensing and reporting channels. To reduce the effects of imperfect channel conditions, a clustering algorithm is suggested in which final decision about the primary user activity is obtained based on linear combination of clusters transmits. To calculate the combination weights, we encounter with the problem of the joint distribution approximation for sum of correlated log-normal variables. A joint MGF matching algorithm is proposed to estimate the sums by a single log-normal vector. Monte Carlo simulations confirm the accuracy of the proposed MGF-based approach and efficiency of cluster based spectrum sensing algorithm in terms of primary signal detection.