LogDet Covariance Based Spectrum Sensing Under Colored Noise

Abstract

Cognitive radio is proposed for efficient utilization of radio spectrum using dynamic spectrum allocation. Most of the spectrum sensing algorithms for cognitive radio are based on some operations on eigenvalue of the covariance matrix of received symbols. However, the performance of existing algorithms degrade in the presence of colored noise due to swelling effect of the covariance matrix. To circumvent this limitation, we propose to use LogDet of covariance matrix of the received samples for spectrum sensing in cognitive radio. The proposed algorithms work extremely well at low signal-to-noise ratio in the presence of both white and colored noise. We also derive the detection threshold of the proposed algorithms. Simulation results show robustness of the proposed algorithms over existing algorithms on the captured digital television signal.