A gradient descent implementation of the adaptive robust narrowband constrained LMS beamformer

Abstract

Classical adaptive beamforming strategies have been shown to exhibit a substantial performance degradation if an error exists in the estimation of the angle of arrival (AOA). The optimization problem can be modified to incorporate an improved constraint allowing for robustness to AOA imperfections. In this paper, we propose a gradient descent implementation of the robust adaptive beamformer whereby a direct relation is established between the Lagrange multiplier and the adaptation step size. This allows for the heuristic approximation of one of the parameters while establishing a valid range for the other. Computer simulations have been used to confirm the findings of this study and illustrate the performance of the developed algorithm.

http://ift.tt/2y7IxhT