Spectrum sensing in the presence of Cauchy noise through differential entropy

The distribution of the noise process in communication systems is usually taken to be Gaussian, albeit the assumption has limited scope. Noise with heavy-tailed distribution is a reality in wireless communication systems, more often than not. Among the non-Gaussians, the Cauchy distribution captures the heavy-tailed nature quite closely. Thus, it is known as a best-fit to model the impulsive nature of noise, frequented in communication systems. In this work, the problem of spectrum sensing in cognitive radios is considered, with Cauchy noise. The proposed algorithm is based on an estimate of the differential entropy in the received observations; the estimate increases under hypothesis than under the null. Through simulations, the efficacy of the proposed method is demonstrated and is shown to compare favourably with the well-known energy based detector.