High speed least square adaptive filter design using ART and HAT

Author

Chi, Zhipei ; Parhi, Keshab K.

Author_Institution

Dept. of Electr. & Comput. Eng., Minnesota Univ., Minneapolis, MN, USA

fYear

2000

fDate

2000

Firstpage

427

Lastpage

436

Abstract

In this paper, annihilation reordering transformation (ART) for pipelining QR decomposition based recursive least square adaptive filters is fully developed and rigorous mathematical proof of ART is given. A novel transformation, referred to as hybrid annihilation transformation (HAT), for pipelining both the QR decomposition based recursive least square adaptive filters and QR decomposition based single/multi-channel least square lattice adaptive filters is presented. HAT does not introduce any approximation in the entire filtering process. Therefore, it causes no performance degradation no matter how deep the filter is pipelined. It allows a linear speedup of the throughput rate by a linear increase of hardware complexity

Keywords

adaptive filters; digital filters; lattice filters; least squares approximations; matrix decomposition; network synthesis; pipeline processing; recursive filters; systolic arrays; ART; HAT; QR decomposition pipelining; QRD-RLS systolic arrays; annihilation reordering transformation; digital filters; hardware complexity; high speed adaptive filter design; hybrid annihilation transformation; linear speedup; omputer simulations; recursive least square adaptive filters; single/multi-channel LS lattice adaptive filters; throughput rate; Adaptive filters; Degradation; Filtering; Hardware; Lattices; Least squares approximation; Least squares methods; Pipeline processing; Subspace constraints; Throughput;

fLanguage

English

Publisher

ieee

Conference_Titel

Signal Processing Systems, 2000. SiPS 2000. 2000 IEEE Workshop on

Conference_Location

Lafayette, LA

ISSN

1520-6130

Print_ISBN

0-7803-6488-0

Type

conf

DOI

10.1109/SIPS.2000.886741

Filename

886741