Modelo Estocástico do Algoritmo NLMS para Sinais de Entrada Gaussianos Complexos

This paper presents the stochastic modeling of a well-known adaptive algorithm from the literature, the normalized least-mean-square (NLMS) algorithm. Considering a system identification problem with stationary plant and complex Gaussian input data, an accurate model obtained analytically is derived here. Such accuracy is mainly due to the form how the calculus of the normalized moments is performed. Simulation results for different operation scenarios are presented, showing that the proposed model predicts satisfactorily the algorithm behavior for both transient and steady-state phases.