TRACKING PROPERTIES OF ADAPTIVE SIGNAL PROCESSING ALGORITHMS.

David C. Farden, Khalid Sayood

Research output: Contribution to conferencePaper

8 Citations (Scopus)

Abstract

Adaptive signal processing algorithms are often used in order to ″track″ an unknown time-varying parameter vector. Such algorithms are typically some form of stochastic gradient-descent algorithm. The Widrow LMS algorithm is apparently the most frequently used. This work develops an upper bound on the norm-squared error between the parameter vector being tracked and the value obtained by the algorithm. The upper bound illustrates the relationship between the algorithm step-size and the maximum rate of variation in the parameter vector. Finally, some simple covariance decay-rate conditions are imposed to obtain a bound on the mean square error.

Original languageEnglish (US)
Pages466-469
Number of pages4
StatePublished - Jan 1 1980
EventUnknown conference - Denver, CO, USA
Duration: Apr 9 1980Apr 11 1980

Other

OtherUnknown conference
CityDenver, CO, USA
Period4/9/804/11/80

Fingerprint

Signal processing
Mean square error

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Farden, D. C., & Sayood, K. (1980). TRACKING PROPERTIES OF ADAPTIVE SIGNAL PROCESSING ALGORITHMS.. 466-469. Paper presented at Unknown conference, Denver, CO, USA, .

TRACKING PROPERTIES OF ADAPTIVE SIGNAL PROCESSING ALGORITHMS. / Farden, David C.; Sayood, Khalid.

1980. 466-469 Paper presented at Unknown conference, Denver, CO, USA, .

Research output: Contribution to conferencePaper

Farden, DC & Sayood, K 1980, 'TRACKING PROPERTIES OF ADAPTIVE SIGNAL PROCESSING ALGORITHMS.' Paper presented at Unknown conference, Denver, CO, USA, 4/9/80 - 4/11/80, pp. 466-469.
Farden DC, Sayood K. TRACKING PROPERTIES OF ADAPTIVE SIGNAL PROCESSING ALGORITHMS.. 1980. Paper presented at Unknown conference, Denver, CO, USA, .
Farden, David C. ; Sayood, Khalid. / TRACKING PROPERTIES OF ADAPTIVE SIGNAL PROCESSING ALGORITHMS. Paper presented at Unknown conference, Denver, CO, USA, .4 p.
@conference{7fb0576a756f45b1bb8db419664ded56,
title = "TRACKING PROPERTIES OF ADAPTIVE SIGNAL PROCESSING ALGORITHMS.",
abstract = "Adaptive signal processing algorithms are often used in order to ″track″ an unknown time-varying parameter vector. Such algorithms are typically some form of stochastic gradient-descent algorithm. The Widrow LMS algorithm is apparently the most frequently used. This work develops an upper bound on the norm-squared error between the parameter vector being tracked and the value obtained by the algorithm. The upper bound illustrates the relationship between the algorithm step-size and the maximum rate of variation in the parameter vector. Finally, some simple covariance decay-rate conditions are imposed to obtain a bound on the mean square error.",
author = "Farden, {David C.} and Khalid Sayood",
year = "1980",
month = "1",
day = "1",
language = "English (US)",
pages = "466--469",
note = "Unknown conference ; Conference date: 09-04-1980 Through 11-04-1980",

}

TY - CONF

T1 - TRACKING PROPERTIES OF ADAPTIVE SIGNAL PROCESSING ALGORITHMS.

AU - Farden, David C.

AU - Sayood, Khalid

PY - 1980/1/1

Y1 - 1980/1/1

N2 - Adaptive signal processing algorithms are often used in order to ″track″ an unknown time-varying parameter vector. Such algorithms are typically some form of stochastic gradient-descent algorithm. The Widrow LMS algorithm is apparently the most frequently used. This work develops an upper bound on the norm-squared error between the parameter vector being tracked and the value obtained by the algorithm. The upper bound illustrates the relationship between the algorithm step-size and the maximum rate of variation in the parameter vector. Finally, some simple covariance decay-rate conditions are imposed to obtain a bound on the mean square error.

AB - Adaptive signal processing algorithms are often used in order to ″track″ an unknown time-varying parameter vector. Such algorithms are typically some form of stochastic gradient-descent algorithm. The Widrow LMS algorithm is apparently the most frequently used. This work develops an upper bound on the norm-squared error between the parameter vector being tracked and the value obtained by the algorithm. The upper bound illustrates the relationship between the algorithm step-size and the maximum rate of variation in the parameter vector. Finally, some simple covariance decay-rate conditions are imposed to obtain a bound on the mean square error.

UR - http://www.scopus.com/inward/record.url?scp=0019210597&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0019210597&partnerID=8YFLogxK

M3 - Paper

SP - 466

EP - 469

ER -