A new algorithm for updating adaptive system

Author

Sanubari, J. ; Tokuda, Keiichi

Author_Institution

Dept. of Electron. Eng., Satya Wacana Univ., Salatiga, Indonesia

Volume

1

fYear

2002

fDate

2002

Firstpage

393

Abstract

This paper presents a new cost function for updating adaptive filter parameters. We utilize a nonlinear cost function that assigns a small step size for large errors and a large step size when the error is small, in order to get fast convergence and small steady state misadjustment. The cost function is derived by assuming that the error is t(λ) distributed. The convergence and the misadjustment mean square error of the proposed method adaptive system output are compared with those of the conventional least square algorithm.

Keywords

adaptive filters; convergence of numerical methods; error statistics; mean square error methods; LMS; adaptive filter parameter updating algorithms; adaptive system output misadjustment mean square error; convergence; error dependent nonlinear cost function step size; error distribution; least square algorithms; steady state misadjustment; Adaptive equalizers; Adaptive filters; Adaptive systems; Convergence; Cost function; Least squares approximation; Least squares methods; Mean square error methods; Steady-state; System identification;

fLanguage

English

Publisher

ieee

Conference_Titel

Circuits and Systems, 2002. APCCAS '02. 2002 Asia-Pacific Conference on

Print_ISBN

0-7803-7690-0

Type

conf

DOI

10.1109/APCCAS.2002.1114978

Filename

1114978