Local spectrum sensing in non-Gaussian noise

This paper deals with the local spectrum sensing problem in non-Gaussian noise. It is of great importance to reliably detect the presence of licensed users in cognitive radio. This task highly depends on the noise distribution, thus it is important to characterize the noise behavior as best as possible. Although Gaussian assumption for noise is of interest due to many reasons, it sometimes fails to model the behavior of noise in practice. In this paper we present a local spectrum sensing scheme in a broad family of non-Gaussian noise. We reasonably assume that the Primary User (PU) signal is Phase Shift Keying (PSK) modulated, where the number of constellation points, M, and the symbol code sequence are unknown to the Secondary Users (SU), as in practice. We also assume that the SUs are unaware of the PU signal complex amplitude and the non-Gaussian noise variance. Within this framework, we obtain the Generalized Likelihood Ratio Test (GLRT) as our spectrum sensing scheme and propose an approximate implementation of the resulting GLRT. Simulation results show that our proposed scheme has an acceptable performance.