An F-Test Based Approach for Spectrum Sensing in Cognitive Radio

Spectrum sensing is a key task in cognitive radio networks. Traditional sensing techniques such as energy detector suffer from noise uncertainty problem or require high computational complexity. In this paper, we propose a novel sensing technique using F-test by considering a multiple antenna cognitive radio system. This method is insensitive to noise uncertainty and easy to implement. It requires the channel state information (CSI) as prior knowledge. Based on statistical properties of F-distribution, we shall derive the test threshold and probability of detection, respectively. In addition, the performance of the proposed approach under imperfect channel information will be discussed. Simulation results show that the proposed F-test based detector achieves significant performance improvement compared with several popular detectors and offers robustness against noise uncertainty.