Ultra low complexity adaptive beamforming via non-eigen decomposition

This paper presents a non-eigendecomposition based beamforming algorithm under the assumption that the power of the desired signal is large compared to the interfering signals. The algorithm provides a noise free suboptimal weight vector when the noise is spatially uncorrelated. and the weight vector is approximately equal to the desired signal´s spatial signature. The significance here is the substantial reduction in computational complexity. Total computational load for the weight vector is O(3M-2) per snapshot for a system with M receiving antennas by approximating the cross correlation vector of the received signals in the reference and other antennas. The weight vector is a function of only the cross correlation vector and initial guess and does not require a step size. This technique requires neither a training sequence nor an assumption of incoherency among impinging signals. The algorithm is derived analytically and simulations evaluate the CDMA capacity improvement and the tracking ability when the incident angle of the desired signal varies at each snapshot.