A fast decision feedback LMS algorithm using multiple step sizes

Author

Modlin, Cory S. ; Cioffi, John M.

Author_Institution

Inf. Syst. Lab., Stanford Univ., CA, USA

fYear

1994

fDate

1-5 May 1994

Firstpage

1201

Abstract

An adaptive decision feedback equalizer (DFE) that uses the least-mean-square (LMS) algorithm may converge slowly or have large excess mean square error. The use of previous decisions in updating the feedforward filter can “whiten” the eigenvalues of the autocorrelation matrix for the feedback filter update without affecting the minimum mean square error making convergence faster and reducing misadjustment. The resultant improvement can be enhanced further by letting the update coefficient constant, μ, be a fixed diagonal matrix

Keywords

adaptive equalisers; decision feedback equalisers; eigenvalues and eigenfunctions; filtering theory; least mean squares methods; matrix algebra; DFE; LMS algorithm; adaptive decision feedback equalizer; autocorrelation matrix; convergence; eigenvalues; feedback filter update; feedforward filter; fixed diagonal matrix; least mean square algorithm; mean square error; minimum mean square error; multiple step sizes; update coefficient constant; Autocorrelation; Convergence; Decision feedback equalizers; Eigenvalues and eigenfunctions; Equations; Information systems; Laboratories; Least squares approximation; Mean square error methods; Nonlinear filters;

fLanguage

English

Publisher

ieee

Conference_Titel

Communications, 1994. ICC '94, SUPERCOMM/ICC '94, Conference Record, 'Serving Humanity Through Communications.' IEEE International Conference on

Conference_Location

New Orleans, LA

Print_ISBN

0-7803-1825-0

Type

conf

DOI

10.1109/ICC.1994.368912

Filename

368912