Outlier detection methods of low SNR nodes for cooperative spectrum sensing

In this paper, we have studied methods of detecting low signal-to-noise ratio (SNR) nodes by outlier detection in cooperative spectrum sensing. In many studies in the field of cooperative sensing, it is assumed that an average received energy among sensing nodes is equal. However, in practical wireless environment, the performance of cooperative sensing with low SNR nodes located underground, indoor, and so on, is severely degraded. The goal of this research is to detect the low SNR nodes and minimizes the impacts caused by these low SNR nodes to the cooperative sensing performance. We propose an outlier detection algorithm to exclude low SNR nodes from cooperative sensing nodes by using past accumulated energy data. Several outlier detection methods are compared, and a suitable method for the low SNR node detection algorithm is selected. In addition, the required length of estimation period for the outlier detection is revealed. Simulation results show that the performance of cooperative sensing is improved even if the low SNR nodes exist in the cooperative nodes.