A Square-Error-Based Regularization for Normalized LMS Algorithms

The purpose of a variable step-size normalized LMS filter is to solve the dilemma of fast convergence or low steady-state error associated with the fixed regularized NLMS. By employing the inverse of weighted square-error as the time-varying regularization parameter, we introduce a new regularization for NLMS algorithms. Extensive simulation results demonstrate that our proposed algorithm outperforms existing schemes in speed of convergence, tracking ability, and low misadjustment.