Robust Spectrum Sensing for Small-Scale Primary Users under Low Signal-to-Noise Ratio

A novel robust detection algorithm based on goodness-of-fit test is proposed to achieve spectrum sensing for the small-scale primary users (SSPUs) including wireless microphones and mobile devices. In the proposed algorithm, a new test statistic is proposed to measure the non-Gaussianity of the power spectral density (PSD) of the received signals. Since the estimated PSD of the ambient Gauss noise theoretically follows a scaled χ2 distribution, spectrum sensing can be achieved by measuring and analyzing how far the estimated PSD of the SU's received signals deviates from the noise's PSD distribution. Moreover, theoretical analysis of decision threshold setting for the proposed algorithm is also discussed. Since the designed test statistic and decision threshold of the proposed algorithm both are independent on any priory knowledge of SSPUs and noise, our algorithm is more robust to the uncertain noise and outperforms some existing spectrum sensing algorithms even under low signal-to-noise ratio (like -30dB).