Fast Spectrum Sensing Using Spatial Higher Order Statistics With Large Receive Antenna Array

A novel spectrum sensing (SS) technique based on the higher order statistics (HOS) of spatial samples, obtained through a large receive antenna array, is proposed. Unlike conventional SS schemes, our approach, relying mainly on large number of spatial samples, exploits the spatial properties of the primary user's signals and differentiates them from the Gaussian noises, white or colored, across different antennas. The novelty of our approach lies in treating the spatial samples as though they are the time-domain samples of a non-stationary random process (NSRP). This, in turn, enables the application of HOS-based detection of NSRP while taking advantage of its asymptotic properties as the number of antennas grows large. Simulation results show that, over one time instant, the proposed method is able to detect primary signals under correlated Gaussian noise and is superior to existing approaches in terms of its robustness against any variation of the correlation coefficient of the noise.