Reducing the effect of finite wordlength on the performance of an LMS adaptive filter

Author

Eweda, Eweda ; Yousef, Nabil ; El-Ramly, Salwa

Volume

2

fYear

1998

fDate

7-11 Jun 1998

Firstpage

688

Abstract

The paper proposes a modification of the LMS algorithm that reduces the effect of finite wordlength on the adaptive filter performance. The proposed algorithm is based on repeatedly freezing the filter coefficients for a certain period of time and then updating them on the base of the average innovation term during the freezing period. Expressions are derived for the steady state mean square error and the convergence time of the proposed algorithm. It is found that the proposed algorithm poses a significantly higher resistance to roundoff errors than that of the conventional LMS algorithm. The derived expressions are found to be in an excellent agreement with computer simulation results

Keywords

adaptive filters; adaptive signal processing; convergence of numerical methods; filtering theory; least mean squares methods; roundoff errors; LMS adaptive filter; LMS algorithm modification; adaptive filter performance; average innovation term; computer simulation results; convergence time; filter coefficients; finite wordlength; freezing period; roundoff errors; steady state mean square error; Adaptive filters; Computer simulation; Convergence; Least squares approximation; Mean square error methods; Proposals; Quantization; Roundoff errors; Steady-state; Technological innovation;

fLanguage

English

Publisher

ieee

Conference_Titel

Communications, 1998. ICC 98. Conference Record. 1998 IEEE International Conference on

Conference_Location

Atlanta, GA

Print_ISBN

0-7803-4788-9

Type

conf

DOI

10.1109/ICC.1998.685098

Filename

685098