Robust broadband beamforming based on frequency invariance constraints and worst-case performance optimization

Two novel broadband beamformers robust against look direction estimation errors are proposed. One is designed by imposing a frequency invariance constraint and limiting the magnitude response of the beamformer within a pre-defined range, and the other one is obtained by combining the frequency invariance constraint and the worst-case performance optimization method, both of which are formulated as convex optimization problems and solved efficiently using existing convex optimisation techniques. Compared with conventional methods, a more efficient and effective control over the beamformer's response is achieved, leading to an improved performance, as shown by simulations.