Fast adaptive algorithms using eigenspace projections

N. Gopalan Nair, Andreas Spanias

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

Although adaptive gradient algorithms are simple and relatively robust, they generally have poor performance in the absence of "rich" excitation. In particular, it is well known that the convergence speed of the LMS algorithm deteriorates when the condition number of the input autocorrelation matrix is large. This problem has been previously addressed using weighted RLS or normalized frequency-domain algorithms. In this paper, we present a new approach that employs gradient projections in selected eigenvector sub-spaces to improve the convergence properties of IAIS algorithms for colored inputs. We also introduce an efficient method to iteratively update an "eigen subspace" of the autocorrelation matrix. The proposed algorithm is more efficient, in terms of computational complexity, than the WRLS and its convergence speed approaches that of the WRLS even for highly correlated inputs.

Original languageEnglish (US)
Title of host publicationConference Record - 28th Asilomar Conference on Signals, Systems and Computers, ACSSC 1994
PublisherIEEE Computer Society
Pages1520-1524
Number of pages5
ISBN (Electronic)0818664053
DOIs
StatePublished - Jan 1 1994
Event28th Asilomar Conference on Signals, Systems and Computers, ACSSC 1994 - Pacific Grove, United States
Duration: Oct 31 1994Nov 2 1994

Publication series

NameConference Record - Asilomar Conference on Signals, Systems and Computers
Volume2
ISSN (Print)1058-6393

Conference

Conference28th Asilomar Conference on Signals, Systems and Computers, ACSSC 1994
CountryUnited States
CityPacific Grove
Period10/31/9411/2/94

Fingerprint

Adaptive algorithms
Autocorrelation
Eigenvalues and eigenfunctions
Computational complexity

ASJC Scopus subject areas

  • Signal Processing
  • Computer Networks and Communications

Cite this

Nair, N. G., & Spanias, A. (1994). Fast adaptive algorithms using eigenspace projections. In Conference Record - 28th Asilomar Conference on Signals, Systems and Computers, ACSSC 1994 (pp. 1520-1524). [471712] (Conference Record - Asilomar Conference on Signals, Systems and Computers; Vol. 2). IEEE Computer Society. https://doi.org/10.1109/ACSSC.1994.471712

Fast adaptive algorithms using eigenspace projections. / Nair, N. Gopalan; Spanias, Andreas.

Conference Record - 28th Asilomar Conference on Signals, Systems and Computers, ACSSC 1994. IEEE Computer Society, 1994. p. 1520-1524 471712 (Conference Record - Asilomar Conference on Signals, Systems and Computers; Vol. 2).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Nair, NG & Spanias, A 1994, Fast adaptive algorithms using eigenspace projections. in Conference Record - 28th Asilomar Conference on Signals, Systems and Computers, ACSSC 1994., 471712, Conference Record - Asilomar Conference on Signals, Systems and Computers, vol. 2, IEEE Computer Society, pp. 1520-1524, 28th Asilomar Conference on Signals, Systems and Computers, ACSSC 1994, Pacific Grove, United States, 10/31/94. https://doi.org/10.1109/ACSSC.1994.471712
Nair NG, Spanias A. Fast adaptive algorithms using eigenspace projections. In Conference Record - 28th Asilomar Conference on Signals, Systems and Computers, ACSSC 1994. IEEE Computer Society. 1994. p. 1520-1524. 471712. (Conference Record - Asilomar Conference on Signals, Systems and Computers). https://doi.org/10.1109/ACSSC.1994.471712
Nair, N. Gopalan ; Spanias, Andreas. / Fast adaptive algorithms using eigenspace projections. Conference Record - 28th Asilomar Conference on Signals, Systems and Computers, ACSSC 1994. IEEE Computer Society, 1994. pp. 1520-1524 (Conference Record - Asilomar Conference on Signals, Systems and Computers).
@inproceedings{791952bcf9cb41b9aafcb3f031cc6233,
title = "Fast adaptive algorithms using eigenspace projections",
abstract = "Although adaptive gradient algorithms are simple and relatively robust, they generally have poor performance in the absence of {"}rich{"} excitation. In particular, it is well known that the convergence speed of the LMS algorithm deteriorates when the condition number of the input autocorrelation matrix is large. This problem has been previously addressed using weighted RLS or normalized frequency-domain algorithms. In this paper, we present a new approach that employs gradient projections in selected eigenvector sub-spaces to improve the convergence properties of IAIS algorithms for colored inputs. We also introduce an efficient method to iteratively update an {"}eigen subspace{"} of the autocorrelation matrix. The proposed algorithm is more efficient, in terms of computational complexity, than the WRLS and its convergence speed approaches that of the WRLS even for highly correlated inputs.",
author = "Nair, {N. Gopalan} and Andreas Spanias",
year = "1994",
month = "1",
day = "1",
doi = "10.1109/ACSSC.1994.471712",
language = "English (US)",
series = "Conference Record - Asilomar Conference on Signals, Systems and Computers",
publisher = "IEEE Computer Society",
pages = "1520--1524",
booktitle = "Conference Record - 28th Asilomar Conference on Signals, Systems and Computers, ACSSC 1994",

}

TY - GEN

T1 - Fast adaptive algorithms using eigenspace projections

AU - Nair, N. Gopalan

AU - Spanias, Andreas

PY - 1994/1/1

Y1 - 1994/1/1

N2 - Although adaptive gradient algorithms are simple and relatively robust, they generally have poor performance in the absence of "rich" excitation. In particular, it is well known that the convergence speed of the LMS algorithm deteriorates when the condition number of the input autocorrelation matrix is large. This problem has been previously addressed using weighted RLS or normalized frequency-domain algorithms. In this paper, we present a new approach that employs gradient projections in selected eigenvector sub-spaces to improve the convergence properties of IAIS algorithms for colored inputs. We also introduce an efficient method to iteratively update an "eigen subspace" of the autocorrelation matrix. The proposed algorithm is more efficient, in terms of computational complexity, than the WRLS and its convergence speed approaches that of the WRLS even for highly correlated inputs.

AB - Although adaptive gradient algorithms are simple and relatively robust, they generally have poor performance in the absence of "rich" excitation. In particular, it is well known that the convergence speed of the LMS algorithm deteriorates when the condition number of the input autocorrelation matrix is large. This problem has been previously addressed using weighted RLS or normalized frequency-domain algorithms. In this paper, we present a new approach that employs gradient projections in selected eigenvector sub-spaces to improve the convergence properties of IAIS algorithms for colored inputs. We also introduce an efficient method to iteratively update an "eigen subspace" of the autocorrelation matrix. The proposed algorithm is more efficient, in terms of computational complexity, than the WRLS and its convergence speed approaches that of the WRLS even for highly correlated inputs.

UR - http://www.scopus.com/inward/record.url?scp=46949097591&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=46949097591&partnerID=8YFLogxK

U2 - 10.1109/ACSSC.1994.471712

DO - 10.1109/ACSSC.1994.471712

M3 - Conference contribution

AN - SCOPUS:46949097591

T3 - Conference Record - Asilomar Conference on Signals, Systems and Computers

SP - 1520

EP - 1524

BT - Conference Record - 28th Asilomar Conference on Signals, Systems and Computers, ACSSC 1994

PB - IEEE Computer Society

ER -