A normalized block LMS algorithm for frequency-domain Volterra filters

Author

Im, Sungbin

Author_Institution

Dept. of inf. & Telecommun. Eng., Soongsil Univ., Seoul, South Korea

fYear

1997

fDate

21-23 Jul 1997

Firstpage

152

Lastpage

156

Abstract

The objective of the paper is to introduce a new adaptive filtering algorithm for estimating frequency-domain second-order Volterra filter coefficients. The approach rests upon the normalized LMS (NLMS) algorithm and the frequency-domain block LMS algorithm. The utilization of the normalized LMS algorithm facilitates choice of a proper step size, with which the adaptive frequency domain Volterra filter is guaranteed to be convergent in the mean-squared sense, and improves convergence rate. The frequency-domain block LMS algorithm estimates frequency-domain second-order Volterra filter coefficients which correspond to the DFT of the time-domain Volterra filter coefficients

Keywords

adaptive filters; convergence of numerical methods; estimation theory; frequency-domain analysis; least mean squares methods; nonlinear filters; adaptive filtering algorithm; block LMS algorithm; convergence rate; frequency-domain Volterra filters; normalized block LMS algorithm; second-order Volterra filter coefficients; step size; Adaptive filters; Convergence; Digital filters; Frequency domain analysis; Frequency estimation; Information filtering; Information filters; Least squares approximation; Nonlinear filters; Time domain analysis;

fLanguage

English

Publisher

ieee

Conference_Titel

Higher-Order Statistics, 1997., Proceedings of the IEEE Signal Processing Workshop on

Conference_Location

Banff, Alta.

Print_ISBN

0-8186-8005-9

Type

conf

DOI

10.1109/HOST.1997.613506

Filename

613506