Robust Adaptive Minimum Entropy Beamformer in Impulsive Noise

This paper considers the problem of adaptive beamforming in alpha-stable (non-Gaussian) noise using the constrained minimum output entropy (MOE) based algorithm. Following the same rational that lead to the least mean p-norm (LMP), the weight update adjustment for minimum output entropy is constrained by statistics higher than second order. Also, the MOE algorithm is very robust to impulsive noise due to its M-estimator property derived from the fact that MOE constrains the output entropy. We explain these results analytically, and through simulations.