Comparison of beamforming algorithms via worst-case SINR maximization

Abstract

In communication systems with interference, the performance of traditional adaptive beamforming to search the direction of angle (DOA) of incoming desired signal degrades because of mismatch between the true array response and the assumed array response of desired signal in a receiver antenna array. In this paper, we approach the method of maximum output Signal-to-Interference-plus-Noise Ratio (SINR) to find array response of the incoming desired signal angle. General method to find the array response with output maximum SINR is an eigen decomposition algorithm. The Eigen decomposition algorithm is derived from eigen decomposition of covariance matrix of incoming signal in a receiver antenna array and find the eigenvector with the largest eigenvalue. Also, the Newton's algorithm uses the gradient vector and Hessian matrix of SINR equation, which formulates of the covariance matrix with incoming signal in a receiver antenna array, and updates the array response. To reduce the complexity computation of Newton's method, the LMS algorithm does not require the Hessian matrix of SINR equation, and updates the array response by using only the gradient vector of SINR equation. We compare the complexity computation and simulate the performances of each algorithm.