Neural blind beamformer for cyclostationary signals

Abstract

A blind beamforming algorithm based on a neural network is presented according to the characteristic of cyclostationary signals. This method transforms the question of estimating beamformer weight vectors into the one of computing the SVD of the cross correlation matrix of array input signals and their frequency shift signals. A cross correlation neural network is introduced to compute the SVD of the cross correlation matrix so as to reduce the computational complexity and carry out the blind beamforming more efficiently. The improved cross-coupled Hebbian learning rule presented can make the weights of the neural network converge much fast. Therefore, it is more promising in the practical use. This method can restrain noise and interference. Simulation proves its correctness.