Adaptive uncertainty based iterative robust capon beamformer

This paper proposes an iterative robust capon beamformer (IRCB) with adaptive uncertainty level. The approach iteratively estimates the actual steering vector based on conventional RCB formulation. Instead of using a fixed uncertainty level, we adaptively update the uncertainty level at each iteration. The uncertainty is updated based on the projection of the presumed steering vector onto the noise subspace. The iteration converges when the noise subspace projection is zero. We additionally impose a constraint that restricts the estimated steering vector to be within a spatial sector in order to overcome inappropriate updating of the uncertainty due to subspace swap at low signal-to-noise ratio (SNR) case. Simulation results show that the adaptive uncertainty based IRCB converges faster than the fixed uncertainty based IRCB. Besides alleviating the need to assume any knowledge on the mismatch, the proposed beamformer requires lower computational cost as compared to the convex optimization based beamformer design.