Further results on multiantenna spectrum sensing in colored noise

The problem of multiantenna spectrum sensing is considered where the receiver noise is allowed to be temporally colored with unknown power spectral density (PSD), but must be spatially uncorrelated. The signal is received over a possibly frequency-selective, unknown channel. A generalized likelihood ratio test (GLRT) for multiantenna spectrum sensing in colored noise under this set-up was recently proposed in the literature based on analysis of the multivariate PSD of the received multiantenna signal. In this paper we analyze the performance (probability of detection) of this GLRT by deriving the asymptotic distribution of the test statistic under the alternative hypothesis. Computer simulation results show an excellent agreement with theory in calculating the probability of detection for a specified probability of false alarm.