Hardware Efficient Adaptive Beamformer Based on Cyclic Variable Step Size

This paper presents an improved adaptive beamforming solution based on Global Sidelobe Canceler (GSC) and Least Mean Square (LMS). By analyzing the impact of step size on LMS performance, a simple yet effective new Cyclic Variable Step Size (CVSS) algorithm is proposed to enhance the convergence speed and excess MSE performance of LMS. By adding just a few shift operations to adapt the step size cyclically, the CVSS LMS is efficient for hardware implementation. Initial study shows that it's possible to double the convergence speed with minimum overhead. The numerical simulation confirmed the effectiveness and performance of the algorithm.