On sequential estimation of linear models from data with correlated noise

Yunlong Wang, Petar M. Djurić

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

Abstract

In this paper, we consider the problem of Bayesian sequential estimation on a set of time invariant parameters. At every time instant, a new observation through a linear model is obtained where the observations are distorted by spatially correlated noise with unknown covariance, whereas in time, the noise samples are independent and identically distributed. We derive the joint posterior of the parameters of interest and the covariance, and we propose several approximations to make the Bayesian estimation tractable. Then we propose a method for forming a pseudo posterior, which is suitable for settings where estimation over networks is applied. By computer simulations, we demonstrate that the Kullback-Leibler divergence between the pseudo posterior and a posterior obtained from a known covariance decreases as the acquisition of new observations continues. We also provide computer simulations that compare the proposed method with the least squares method.

Original languageEnglish (US)
Title of host publication2014 Proceedings of the 22nd European Signal Processing Conference, EUSIPCO 2014
PublisherEuropean Signal Processing Conference, EUSIPCO
Pages2365-2369
Number of pages5
ISBN (Electronic)9780992862619
StatePublished - Nov 10 2014
Event22nd European Signal Processing Conference, EUSIPCO 2014 - Lisbon, Portugal
Duration: Sep 1 2014Sep 5 2014

Publication series

NameEuropean Signal Processing Conference
ISSN (Print)2219-5491

Other

Other22nd European Signal Processing Conference, EUSIPCO 2014
CountryPortugal
CityLisbon
Period9/1/149/5/14

Keywords

  • Bayesian inference
  • distributed estimation
  • pseudo posterior
  • unknown covariance

Fingerprint Dive into the research topics of 'On sequential estimation of linear models from data with correlated noise'. Together they form a unique fingerprint.

Cite this