Excitation Conditions for Signed Regressor Least Mean Squares Adaptation

W. A. Sethares; Iven M.Y. Mareels; Brian D.O. Anderson; C. Richard Johnson; Robert R. Bitmead

doi:10.1109/31.1799

Excitation Conditions for Signed Regressor Least Mean Squares Adaptation

W. A. Sethares, Iven M.Y. Mareels, Brian D.O. Anderson, C. Richard Johnson, Robert R. Bitmead

School of Engineering

Research output: Contribution to journal › Article › peer-review

56 Citations (Scopus)

Abstract

The stability of the signed regressor variant of least mean square (LMS) adaptation is found to be heavily dependent on the characteristics of the input sequence. Averaging theory is used to derive a persistence of excitation condition which guarantees exponential stability of the signed regressor algorithm. Failure to meet this condition (which is not equivalent to persistent excitation for LMS) can result in exponential instability, even with the use of leakage. This new persistence of excitation condition is then interpreted in both deterministic and stochastic settings.

Original language	English
Pages (from-to)	613-624
Number of pages	12
Journal	IEEE Transactions on Circuits and Systems
Volume	35
Issue number	6
DOIs	https://doi.org/10.1109/31.1799
Publication status	Published - Jun 1988

Access to Document

10.1109/31.1799

Cite this

@article{b0f3b5e5dd014cd8b7a9a9ad37e1a0ca,

title = "Excitation Conditions for Signed Regressor Least Mean Squares Adaptation",

abstract = "The stability of the signed regressor variant of least mean square (LMS) adaptation is found to be heavily dependent on the characteristics of the input sequence. Averaging theory is used to derive a persistence of excitation condition which guarantees exponential stability of the signed regressor algorithm. Failure to meet this condition (which is not equivalent to persistent excitation for LMS) can result in exponential instability, even with the use of leakage. This new persistence of excitation condition is then interpreted in both deterministic and stochastic settings.",

author = "Sethares, {W. A.} and Mareels, {Iven M.Y.} and Anderson, {Brian D.O.} and {Richard Johnson}, C. and Bitmead, {Robert R.}",

year = "1988",

month = jun,

doi = "10.1109/31.1799",

language = "English",

volume = "35",

pages = "613--624",

journal = "IEEE Transactions on Circuits and Systems",

issn = "0098-4094",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "6",

}

TY - JOUR

T1 - Excitation Conditions for Signed Regressor Least Mean Squares Adaptation

AU - Sethares, W. A.

AU - Mareels, Iven M.Y.

AU - Anderson, Brian D.O.

AU - Richard Johnson, C.

AU - Bitmead, Robert R.

PY - 1988/6

Y1 - 1988/6

N2 - The stability of the signed regressor variant of least mean square (LMS) adaptation is found to be heavily dependent on the characteristics of the input sequence. Averaging theory is used to derive a persistence of excitation condition which guarantees exponential stability of the signed regressor algorithm. Failure to meet this condition (which is not equivalent to persistent excitation for LMS) can result in exponential instability, even with the use of leakage. This new persistence of excitation condition is then interpreted in both deterministic and stochastic settings.

AB - The stability of the signed regressor variant of least mean square (LMS) adaptation is found to be heavily dependent on the characteristics of the input sequence. Averaging theory is used to derive a persistence of excitation condition which guarantees exponential stability of the signed regressor algorithm. Failure to meet this condition (which is not equivalent to persistent excitation for LMS) can result in exponential instability, even with the use of leakage. This new persistence of excitation condition is then interpreted in both deterministic and stochastic settings.

UR - http://www.scopus.com/inward/record.url?scp=0024033701&partnerID=8YFLogxK

U2 - 10.1109/31.1799

DO - 10.1109/31.1799

M3 - Article

AN - SCOPUS:0024033701

SN - 0098-4094

VL - 35

SP - 613

EP - 624

JO - IEEE Transactions on Circuits and Systems

JF - IEEE Transactions on Circuits and Systems

IS - 6

ER -

Excitation Conditions for Signed Regressor Least Mean Squares Adaptation

Abstract

Access to Document

Other files and links

Fingerprint

Cite this