Signal-to-Noise Ratio (SNR) estimation in additive Gaussian mixture noise channel

SNR estimation has been studied extensively in the past. Nevertheless, vast majority of prior works in the design of SNR estimation algorithms are mainly focused on the assumption of Gaussian noise models. It is often assumed that the receiver noise is Gaussian distributed and arises from the receiving system itself. However, the type of noise commonly encountered in practice is reported to be non Gaussian due to man-made noise and interference. As a consequence, Gaussian based SNR estimator performs poorly when the distribution of the noise deviates from Gaussian. This study aims to investigate the efficiency and robustness of the existing Gaussian-based SNR estimators when the noise distribution deviates from Gaussian and also to design an optimum SNR estimator for non-Gaussian noise channel. The two-term additive Gaussian mixture noise (AGMN) is adopted to model the non Gaussian noise. Simulation results show that the performance of existing Gaussian based SNR estimators degrades in the AGMN channel. Hence, the design of a robust SNR estimator in AGMN is necessary