Clustering formation in cognitive radio networks using machine learning

The goal of spectrum sensing is to elevate the detection performance of secondary users (SUs) in a cognitive radio network (CRN). In cooperative spectrum sensing, all secondary users (SUs) of the network deliver their sensing measurement to the fusion center (FC) for the final decision regarding the activity of primary user (PU). The collaboration among large number of SUs might create overhead for the FC. To improve the performance of cooperative spectrum sensing, a novel method is proposed, which segregates the network into clusters. We have used artificial intelligence to make the clusters. The formation of clusters is made based on machine learning affinity propagation algorithm. Using proposed method, SUs share local messages with their neighbors until a highest class of cluster heads are chosen and a corresponding clustering configuration is made. The messages are evaluated depend on measures of similarity between the SUs, which are selected based on the objective of the clustering process. The sensing message of delimited number of SUs is shared with their cluster heads, which is ultimately shared with the FC for final decision. The proposed approach obtains the highest energy and performance efficiency in comparison with conventional clustering schemes.