A new variable step size LMS adaptive algorithm

Abstract

LMS adaptive filtering algorithm is widely used in adaptive control system. To a certain extent the variable step size LMS algorithm can solve the conflict between convergence rate and steady-state error. It is difficult for the traditional LMS to solve. A new variable step size LMS adaptive algorithm based on the existing algorithm is proposed in this paper. By using the gradient of the filter coefficient vector W(n) the algorithm accelerates the convergence speed on the basis of ensuring the convergence accuracy. At the same time, the update formula of step size is adjusted to enhance the ability of the algorithm to resist noise interference. Finally, the MATLAB simulation results show that the algorithm has faster convergence speed, smaller steady-state error and better robustness.