Adaptive beamforming using affine projection algorithms

The dynamic adaptive linear constrained minimum variance (LCMV) beamformer generally uses the normalized LMS (NLMS) algorithm for its robustness and simplicity, but it suffers from a low convergence rate. This paper proposes an alternative beamformer incorporating a simplified fast affine projection (FAP) algorithm, which provides fast convergence speed and low computational complexity. The family of affine projection (AP) algorithms improves its convergence speed significantly due to its decorrelation property. We compare the decorrelation properly of the AP algorithms to that of the spatial smoothing algorithm. The fundamental difference between the two decorrelation properties is clarified.