Finite-Sample Size Multiple Antennas Spectrum Sensing

In this paper, we consider the problem of multiple antenna spectrum sensing in Cognitive Radios (CR) by exploiting the prior information about unknown parameters. Specifically, we consider a blind spectrum sensing problem when the channel gains and the noise variance are unknown for the Secondary User (SU). Under assumption that additional statistical side-information is available about unknown parameters, we use a novel Generalized Likelihood Ratio (GLR) test, which is optimal under finite number of samples, in order to derive our proposed detector. As it has been shown, this novel GLR test need to obtain the Maximum A-posteriori Probability (MAP) estimation of unknown parameters which it is impossible to obtain them in closed form for our case. Thus, we calculate them based on the Expectation-Maximization (EM) algorithm. The simulation results show that our proposed detector has good performance even for finite number of samples and also outperforms the classical GLR detector.