An improved variable-step NLMS algorithm

Aiming at the contradiction between convergence speed and steady state error of the traditional NLMS,an improved variable-step NLMS algorithm is proposed,in which the functional relationship between step-length and error is established,which makes dynamic update of step length changed with variation of output error and noise error to reduce steady-state error. The theoretical analysis and simulation results show that,compared with the existing NLMS algorithm,the improved algorithm has faster convergence speed and lower steady-state error.