Collaborative Spectrum Sensing Based on Signal Correlation in Cognitive Radio Networks

Collaborative spectrum sensing (CSS) attracts great attention due to its advantages to achieve high sensing performance on high reliability, low power consumption in cognitive radio networks. To achieve high accurate performance, recent CSS algorithms based on signal correlation sensing require a large samples. Thus, it is a critical issue for achieving fast and accurate performance simultaneously in CSS. In this paper, we design a fast and highly accurate CSS algorithm based on the sampling correlation matrix calculated from a limited number of received pairwise signal samples. We present a novel alternative testing (AT) method to set the detection threshold dynamically. The AT method is fully blind without requiring the knowledge of signal, channel, noise power, and eigenvalue distribution of correlation matrix. Simulation results show that our proposed scheme achieves high detection probability, low false alarm probability with relative small signal samples.