A variable regularization control method for NLMS algorithm

It is known that regularization plays an important part in adaptive filtering. Several time-varying regularized normalized least-mean-square (NLMS) algorithms have been derived in the past decade. This paper proposes a variable regularization control method for the NLMS algorithm that employs the input signal power, the mean-square error and the estimated system noise power to control the variable regularization parameter. Simulation experiments show that the proposed algorithm performs with fast convergence rate, good tracking, and low misadjustment. Furthermore, the theoretical steady-state behavior is in very good agreement with the experimental results.