Improved Adaptive Beamforming Algorithms for Wireless Systems

The classical least mean square (LMS) algorithm is a widely studied method for adaptive beamforming. It is well known for its lower computational complexity. However, fast and robust beamforming is not possible with the classical LMS method since it uses a constant step size. This nature hinders its applications in many advanced communication systems. Furthermore, this method degrades when the signal-to-noise ratio is rapidly changing. To circumvent these issues posed by the classical LMS method, two modified LMS beamformers are presented in this paper. We name these methods as M-LMS-1 and M-LMS-2. We present two new complex array weights to accelerate the rate of convergence. Computer simulations show that both methods present fast and robust beamforming. That is these algorithms have convergence improvement of about 37.5% and 50% over the standard LMS algorithm.