Distributed bias-compensated recursive least squares estimation over multi-agent networks

  • Jian Lou
  • , Lijuan Jia*
  • , Yangzhi Ye
  • , Zijiang Yang
  • *Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

We study the problem of distributed parameter estimation over multi-agent networks, where agents collaborate to estimate a common parameter vector. Considering the situation where both the input and the output of each agent are corrupted by additive noise, traditional recursive least-squares (RLS) algorithms are biased. The stand-alone bias-compensated RLS (BCRLS) algorithm can remove the effect of noise-induced bias but increases the variance of the local estimator. It has been discussed that the variance can be significantly reduced by distributed cooperation. Therefore, a series of distributed BCRLS algorithms are proposed based on current mainstream distributed cooperation schemes (incremental, consensus and diffusion schemes). And performance of the proposed distributed algorithms is compared. Simulation results show that distributed BCRLS estimation performs better than the noncooperative BCRLS (Nco-BCRLS) estimation. And the diffusion BCRLS (Diff-BCRLS) algorithm has the best performance in estimation accuracy and tracking capability among the distributed algorithms.

Original languageEnglish
Title of host publicationProceedings of the 35th Chinese Control Conference, CCC 2016
EditorsJie Chen, Qianchuan Zhao, Jie Chen
PublisherIEEE Computer Society
Pages7996-8001
Number of pages6
ISBN (Electronic)9789881563910
DOIs
Publication statusPublished - 26 Aug 2016
Event35th Chinese Control Conference, CCC 2016 - Chengdu, China
Duration: 27 Jul 201629 Jul 2016

Publication series

NameChinese Control Conference, CCC
Volume2016-August
ISSN (Print)1934-1768
ISSN (Electronic)2161-2927

Conference

Conference35th Chinese Control Conference, CCC 2016
Country/TerritoryChina
CityChengdu
Period27/07/1629/07/16

Keywords

  • Bias Compensation
  • Distributed Parameter Estimation
  • Distributed Strategies
  • Least-Squares
  • Multi-Agent Networks

Fingerprint

Dive into the research topics of 'Distributed bias-compensated recursive least squares estimation over multi-agent networks'. Together they form a unique fingerprint.

Cite this