A recursive algorithm for robust wideband adaptive beamforming

Abstract

A novel computationally efficient wideband adaptive beamformer, robust against look direction errors, is proposed in this paper. The problem is formulated as Linearly Constrained Minimum Variance Beamformer (LCMV), with the addition of frequency invariance and derivative constraints. The frequency invariance constraint leads to consistent beamforming performance for a wideband of signals whereas derivative constraints ensure the robustness against steering vector mismatch. The gradient descent formulation is used to provide a recursive solution to this convex optimization problem. This results in a robust broadband adaptive beamforming algorithm with low computational complexity compared to the existing robust wide-band adaptive beamforming algorithms. Computer simulations demonstrate comparable performance of proposed algorithm as against traditional LCMV adaptive beamforming algorithm across a wide band of operation, with improved robustness to steering vector mismatches.