Stochastic model for the modified filtered-error LMS algorithm

This paper proposes an improved stochastic model for the first and second moments of the modified filtered-error least-mean-square (MFELMS) algorithm. The proposed model is obtained not invoking the classic Independence Theory (IT), allowing for a slow adaptation assumption and Gaussian input signal. Numerical simulations corroborate the good agreement between the results obtained with the Monte Carlo (MC) method and through the proposed model for both white and colored inputs.