A Fast Beamforming Algorithm for Large Arrays

Abstract

This beamforming algorithm is written specifically for array radars in which the number of array elements K is very large compared with the number of jammers L the radar is designed to suppress. It uses a set of M noise vectors to construct a basis for the jammer component of the antenna output vectors. The component of the quiescent weight vector orthogonal to each basis vector is calculated, renormalized to unit length, and identified as the adapted weight vector. This algorithm is effective in the suppression of many types of jammers. The number of noise samples M required in the construction of the adapted weight vector is approximately equal to L. In the special case of L narrowband noise jammers, for example, a choice of M = L usually reduces the receiver output jammer power to a few dBs above the white noise background. It is permissible to have M<L. In this case, the first M strongest jammers are given the most suppression. This algorithm is very simple in design and programming and requires approximately (4M2 + 6M + 2)K real add and (4M2 + 8M + 4)K real multiply operations. Most of these operations are in the form of vector operations and can be carried out efficiently with a vector array processor.