An Adaptive Spectrum-Sensing Algorithm for Cognitive Radio Networks based on the Sample Covariance Matrix

Abstract

<p>A novel adaptive threshold spectrum sensing technique based on the covariance matrix of received signal samples is proposed. The adaptive threshold in terms of signal to noise ratio (SNR) and spectrum utilisation ratio of primary user is derived. It considers both the probability of detection and the probability false alarm to minimise the overall decision error probability. The energy- based spectrum sensing scheme shows high vulnerability under noise uncertainty and low SNR. The existing covariance-based spectrum sensing technique overcomes the noise uncertainty problem but its performance deteriorates under low SNR. The proposed covariance-based scheme effectively addresses the low SNR problem. The superior performance of this scheme over the existing covariance-based detection method is confirmed by the simulation results in terms of probability of detection, probability of error, and requirement of samples for reliable detection of spectrum.</p>