A fast and robust adaptive beamformer

Abstract

Based on a unitary transformation, improved adaptive beamforming via orthogonal projection is proposed. The new algorithm firstly transforms a complex-valued covariance matrix into a real-valued matrix by means of a unitary transformation, then eigen-decomposes the transformed matrix for adaptive beamforming. The overall computational load can be significantly reduced, to about only one-fourth of that of the original orthogonal projection method. During the course of the computation for the real-valued matrix, the inherent forward-backward averaging effect, which is equivalent to double the number of snapshots, may upgrade the robustness in the case of a small number of snapshots and closely spaced jamming sources and may raise the output signal-to-interference-plus-noise ratio. Additionally, the spatial smoothing can decorrelate possibly correlated source pairs; therefore, the presented method has a better performance for jammer suppression and stronger ability to reshape the beam as compared to the orthogonal projection and sample matrix inversion algorithms in scenarios with partially correlated or fully coherent sources. The performance of the presented algorithm does not depend on the particular choice of the unitary matrix. Computer simulations demonstrate the effectiveness of the proposed method.