On robust spectrum sensing using M-estimators of covariance matrix

Abstract

Most of the spectrum sensing techniques are designed for Gaussian noise. These techniques do not consider the environment with the non-Gaussian (impulsive or heavy-tailed) noise. In a wireless communication system, impulsive noise frequently occurs and originates from numerous sources, for instance, switching transients in power lines, vehicle ignition, microwave ovens and devices with electromechanical switches. Under those circumstances, sensing techniques designed for Gaussian noise may be highly susceptible to severe degradation of performance.