Colored loading for robust adaptive beamforming with low sample support

Abstract

Classical robust adaptive beamforming makes use of diagonal loading to prevent high sidelobes and distorted main beams caused by the inaccurate estimation of the covariance ma- trix, signal mismatch and non-stationary interference. This paper develops a general loading technique, called colored loading (CL), by which the estimated covariance matrix is loaded by a scaled version of a Hermitian matrix, usually a Toeplitz matrix for linear equispaced sensors, as opposed to using the identity matrix as in diagonal loading. The colored loading retains the simplicity of implementation of diagonal loading. Numerical examples show that colored loading is superior to diagonal loading in terms of the optimal output signal-to-interference plus noise ratio (SINR), in the snapshot deficient scenario.