A modified LMS algorithm with improved convergence

Author

Rohani, B. ; Chung, K.S.

Author_Institution

Sch. of Electr. & Comput. Sci., Curtin Univ. of Technol., Perth, WA, Australia

Volume

2

fYear

1994

fDate

14-18 Nov 1994

Firstpage

845

Abstract

The weighted step size (WSS) least mean squares (LMS) algorithm utilises non-uniform step sizes for updating an impulse response estimator. This improves the speed of the convergence of this algorithm compared to the LMS algorithm. The additional computational requirement for this algorithm is the extra update of the algorithm step sizes. A quantitative analysis of the WSS-LMS algorithm is presented. This includes the condition for the algorithm convergence and the steady state mean squared error (MSE) analysis of the algorithm. Computer simulation results that establish the validity of the analysis are also presented

Keywords

adaptive signal processing; convergence of numerical methods; least mean squares methods; transient response; MSE analysis; WSS-LMS algorithm; adaptive signal processing algorithms; computational requirement; computer simulation results; convergence; impulse response estimator updating; least mean squares algorithm; modified LMS algorithm; nonuniform step sizes; steady state mean squared error analysis; weighted step size; Algorithm design and analysis; Australia; Computer errors; Computer simulation; Convergence; Finite impulse response filter; Least squares approximation; Sampling methods; Signal processing algorithms; Steady-state;

fLanguage

English

Publisher

ieee

Conference_Titel

Singapore ICCS '94. Conference Proceedings.

Print_ISBN

0-7803-2046-8

Type

conf

DOI

10.1109/ICCS.1994.474153

Filename

474153