Convergence properties of normalized random incremental gradient algorithms for least-squares source localization

Michael Rabbat; Angelia Nedic

doi:10.1109/ACSSC.2012.6489259

Convergence properties of normalized random incremental gradient algorithms for least-squares source localization

Michael Rabbat, Angelia Nedic

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

1 Scopus citations

Abstract

We consider the problem of localizing a single source using received signal strength measurements gathered at a number of sensors. We assume that the measurements follow the standard path loss model and are corrupted by additive white Gaussian noise. Under this model, the maximum likelihood solution to the source localization problem involves solving a non-linear least squares optimization problem. We study the convergence property of a normalized incremental gradient method for solving this problem. Remarkably, despite the fact that the problem is non-convex, the normalized incremental gradient method generates a sequence of iterates which are attracted to the global optimum under some mild conditions.

Original language	English (US)
Title of host publication	Conference Record of the 46th Asilomar Conference on Signals, Systems and Computers, ASILOMAR 2012
Pages	1417-1421
Number of pages	5
DOIs	https://doi.org/10.1109/ACSSC.2012.6489259
State	Published - Dec 1 2012
Externally published	Yes
Event	46th Asilomar Conference on Signals, Systems and Computers, ASILOMAR 2012 - Pacific Grove, CA, United States Duration: Nov 4 2012 → Nov 7 2012

Publication series

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

Other

Other	46th Asilomar Conference on Signals, Systems and Computers, ASILOMAR 2012
Country/Territory	United States
City	Pacific Grove, CA
Period	11/4/12 → 11/7/12

ASJC Scopus subject areas

Signal Processing
Computer Networks and Communications

Access to Document

10.1109/ACSSC.2012.6489259

Cite this

Rabbat, M., & Nedic, A. (2012). Convergence properties of normalized random incremental gradient algorithms for least-squares source localization. In Conference Record of the 46th Asilomar Conference on Signals, Systems and Computers, ASILOMAR 2012 (pp. 1417-1421). Article 6489259 (Conference Record - Asilomar Conference on Signals, Systems and Computers). https://doi.org/10.1109/ACSSC.2012.6489259

Convergence properties of normalized random incremental gradient algorithms for least-squares source localization. / Rabbat, Michael; Nedic, Angelia.
Conference Record of the 46th Asilomar Conference on Signals, Systems and Computers, ASILOMAR 2012. 2012. p. 1417-1421 6489259 (Conference Record - Asilomar Conference on Signals, Systems and Computers).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Rabbat, M & Nedic, A 2012, Convergence properties of normalized random incremental gradient algorithms for least-squares source localization. in Conference Record of the 46th Asilomar Conference on Signals, Systems and Computers, ASILOMAR 2012., 6489259, Conference Record - Asilomar Conference on Signals, Systems and Computers, pp. 1417-1421, 46th Asilomar Conference on Signals, Systems and Computers, ASILOMAR 2012, Pacific Grove, CA, United States, 11/4/12. https://doi.org/10.1109/ACSSC.2012.6489259

Rabbat M, Nedic A. Convergence properties of normalized random incremental gradient algorithms for least-squares source localization. In Conference Record of the 46th Asilomar Conference on Signals, Systems and Computers, ASILOMAR 2012. 2012. p. 1417-1421. 6489259. (Conference Record - Asilomar Conference on Signals, Systems and Computers). doi: 10.1109/ACSSC.2012.6489259

@inproceedings{726237c2b844456993731f92cff44c20,

title = "Convergence properties of normalized random incremental gradient algorithms for least-squares source localization",

abstract = "We consider the problem of localizing a single source using received signal strength measurements gathered at a number of sensors. We assume that the measurements follow the standard path loss model and are corrupted by additive white Gaussian noise. Under this model, the maximum likelihood solution to the source localization problem involves solving a non-linear least squares optimization problem. We study the convergence property of a normalized incremental gradient method for solving this problem. Remarkably, despite the fact that the problem is non-convex, the normalized incremental gradient method generates a sequence of iterates which are attracted to the global optimum under some mild conditions.",

author = "Michael Rabbat and Angelia Nedic",

year = "2012",

month = dec,

day = "1",

doi = "10.1109/ACSSC.2012.6489259",

language = "English (US)",

isbn = "9781467350518",

series = "Conference Record - Asilomar Conference on Signals, Systems and Computers",

pages = "1417--1421",

booktitle = "Conference Record of the 46th Asilomar Conference on Signals, Systems and Computers, ASILOMAR 2012",

note = "46th Asilomar Conference on Signals, Systems and Computers, ASILOMAR 2012 ; Conference date: 04-11-2012 Through 07-11-2012",

}

TY - GEN

T1 - Convergence properties of normalized random incremental gradient algorithms for least-squares source localization

AU - Rabbat, Michael

AU - Nedic, Angelia

PY - 2012/12/1

Y1 - 2012/12/1

N2 - We consider the problem of localizing a single source using received signal strength measurements gathered at a number of sensors. We assume that the measurements follow the standard path loss model and are corrupted by additive white Gaussian noise. Under this model, the maximum likelihood solution to the source localization problem involves solving a non-linear least squares optimization problem. We study the convergence property of a normalized incremental gradient method for solving this problem. Remarkably, despite the fact that the problem is non-convex, the normalized incremental gradient method generates a sequence of iterates which are attracted to the global optimum under some mild conditions.

AB - We consider the problem of localizing a single source using received signal strength measurements gathered at a number of sensors. We assume that the measurements follow the standard path loss model and are corrupted by additive white Gaussian noise. Under this model, the maximum likelihood solution to the source localization problem involves solving a non-linear least squares optimization problem. We study the convergence property of a normalized incremental gradient method for solving this problem. Remarkably, despite the fact that the problem is non-convex, the normalized incremental gradient method generates a sequence of iterates which are attracted to the global optimum under some mild conditions.

UR - http://www.scopus.com/inward/record.url?scp=84876222821&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84876222821&partnerID=8YFLogxK

U2 - 10.1109/ACSSC.2012.6489259

DO - 10.1109/ACSSC.2012.6489259

M3 - Conference contribution

AN - SCOPUS:84876222821

SN - 9781467350518

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

SP - 1417

EP - 1421

BT - Conference Record of the 46th Asilomar Conference on Signals, Systems and Computers, ASILOMAR 2012

T2 - 46th Asilomar Conference on Signals, Systems and Computers, ASILOMAR 2012

Y2 - 4 November 2012 through 7 November 2012

ER -

Convergence properties of normalized random incremental gradient algorithms for least-squares source localization

Abstract

Publication series

Other

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this