Stochastic modeling of the NLMS algorithm for complex Gaussian input data and nonstationary environment

EDUARDO V KUHN; KOLODZIEJ, JAVIER ERNESTO; SEARA, RUI

DIGITAL SIGNAL PROCESSING

ACADEMIC PRESS INC ELSEVIER SCIENCE

Lugar: Amsterdam; Año: 2014 vol. 30 p. 55 - 66

1051-2004

This paper presents a stochastic model for the normalized least-mean-square (NLMS) algorithm operating under nonstationary environment with complex Gaussian input data. Specifically, model expressions describing the algorithm behavior in both transient and steady-state phases are obtained, allowing therefore a better understanding of how the algorithm parameters affect its performance. The proposed approach avoids several approximations commonly used in the modeling of algorithms with normalized step size, thus giving rise to a very accurate analytical model. Such accuracy is mainly due to the strategy used for computing the normalized autocorrelation-like matrices arising from the model development. Through simulation results, the accuracy of the proposed model is assessed for different operating scenarios.