A Novel Cooperative Spectrum Sensing Algorithm Based on Random Matrix Theory

Spectrum sensing plays an important role in cognitive radio (CR) system which is able to detect the primary users (PU) and avoid secondary users´ (SU) interfering with them. However, the traditional sensing schemes such as energy detection, cyclostationary feature detection and matched detection need know the prior knowledge which is practically hard to obtain. Meanwhile, considering the influence of hidden terminals and multipath fading problems on the sensing result, this paper introduces a novel cooperative spectrum sensing algorithm which uses the eigenvalues of sample covariance matrix based on random matrix theory to sense spectrum holes. The proposed scheme does not require any information of the transmitted signal and the noise power. Numerical results show that the proposed scheme can gain higher sensing performance with a few of secondary users and is more robust to the noise uncertainty compared with the conventional sensing schemes.