Local Variance Detection for Multi-Antenna Spectrum Sensing

Efficient and accurate spectrum sensing is an essential part of cognitive radio. In this letter, we propose two local variance methods for multi-antenna spectrum sensing. By calculating the maximum local variance (MLV) and the average local variance (ALV) of the sample covariance matrix (SCM), respectively, we construct the test statistics to decide whether the spectrum is idle or not. Furthermore, we derive the corresponding decision thresholds according to the asymptotic distribution theory. Since our methods require no prior information and only small sample size, they can be applied in various signal detection. Simulation results show that the proposed methods exhibit better performance than the existing techniques.