Doubly constrained robust beamforming method using subspace-associated power components

Abstract

The Capon beamformer can reject interfering signals by minimizing the array output power subject to the linear constraint exploiting a steering vector. When there exist steering vector errors, however, its performance can be severely degraded as the desired signal is misunderstood as interference. A robust beamforming method is presented which is based on part of the array output power associated with a subspace spanned by eigenvectors corresponding to some largest eigenvalues of the received signal covariance matrix. The use of the subspace-associated power component only, rather than all power components, allows us to find the minimum norm solution, which can result in a reduction in the output noise power since it is proportional to the squared norm of the weight vector. The proposed beamformer solves a minimization problem under the norm constraint and the set constraint on the steering vector. Though the constraint functions of the minimization problem are similar to those in the original doubly constrained beamformer (DCB), the cost function is differently defined based on the subspace power such that better robustness can be attained. Simulation shows that the proposed beamformer is superior in signal-to-interference-plus-noise ratio (SINR) to existing ones such as the conventional DCB.