Robust adaptive beamforming based on covariance matrix and new steering vector estimation

Abstract

A new approach to the design of robust adaptive beamforming is introduced. In this approach, first the sample covariance matrix is replaced by the estimated interference-plus-noise covariance matrix which is constructed by exploiting the spatial spectrum distributions of a generalized Hermitian covariance matrix and the sample covariance matrix. Then, a new robust adaptive beamforming approach is developed based on finding a more accurate estimate of the actual steering vector than the available prior. The essence of finding such a steering vector estimate is the construction of the covariance matrix of the signal-of-interest based on the Capon spectrum estimator. The estimator is based on the integration over an angular region in which the desired signal direction is located. In the proposed method, the norm constraints are not required. Furthermore, it avoids solving quadratic convex optimization programs. The effectiveness of the proposed method is demonstrated by numerical results.