Sequential Bayesian prediction and adaptive sampling algorithms for mobile sensor networks

Yunfei Xu, Jongeun Choi, Sarat Dass, Tapabrata Maiti

Research output: Contribution to journalArticle

27 Citations (Scopus)

Abstract

In this technical note, we formulate a fully Bayesian approach for spatio-temporal Gaussian process regression such that multifactorial effects of observations, measurement noise and prior distributions are all correctly incorporated in the predictive distribution. Using discrete prior probabilities and compactly supported kernels, we provide a way to design sequential Bayesian prediction algorithms in which exact predictive distributions can be computed in constant time as the number of observations increases. For a special case, a distributed implementation of sequential Bayesian prediction algorithms has been proposed for mobile sensor networks. An adaptive sampling strategy for mobile sensors, using the maximum a posteriori (MAP) estimation, has been proposed to minimize the prediction error variances. Simulation results illustrate the practical usefulness of the proposed theoretically-correct algorithms.

Original languageEnglish
Article number6099560
Pages (from-to)2078-2084
Number of pages7
JournalIEEE Transactions on Automatic Control
Volume57
Issue number8
DOIs
Publication statusPublished - 2012 Aug 8

Fingerprint

Sensor networks
Wireless networks
Sampling
Sensors

All Science Journal Classification (ASJC) codes

  • Control and Systems Engineering
  • Computer Science Applications
  • Electrical and Electronic Engineering

Cite this

@article{0c0e4b465c8c421f85fb8f9228a16029,
title = "Sequential Bayesian prediction and adaptive sampling algorithms for mobile sensor networks",
abstract = "In this technical note, we formulate a fully Bayesian approach for spatio-temporal Gaussian process regression such that multifactorial effects of observations, measurement noise and prior distributions are all correctly incorporated in the predictive distribution. Using discrete prior probabilities and compactly supported kernels, we provide a way to design sequential Bayesian prediction algorithms in which exact predictive distributions can be computed in constant time as the number of observations increases. For a special case, a distributed implementation of sequential Bayesian prediction algorithms has been proposed for mobile sensor networks. An adaptive sampling strategy for mobile sensors, using the maximum a posteriori (MAP) estimation, has been proposed to minimize the prediction error variances. Simulation results illustrate the practical usefulness of the proposed theoretically-correct algorithms.",
author = "Yunfei Xu and Jongeun Choi and Sarat Dass and Tapabrata Maiti",
year = "2012",
month = "8",
day = "8",
doi = "10.1109/TAC.2011.2179430",
language = "English",
volume = "57",
pages = "2078--2084",
journal = "IEEE Transactions on Automatic Control",
issn = "0018-9286",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "8",

}

Sequential Bayesian prediction and adaptive sampling algorithms for mobile sensor networks. / Xu, Yunfei; Choi, Jongeun; Dass, Sarat; Maiti, Tapabrata.

In: IEEE Transactions on Automatic Control, Vol. 57, No. 8, 6099560, 08.08.2012, p. 2078-2084.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Sequential Bayesian prediction and adaptive sampling algorithms for mobile sensor networks

AU - Xu, Yunfei

AU - Choi, Jongeun

AU - Dass, Sarat

AU - Maiti, Tapabrata

PY - 2012/8/8

Y1 - 2012/8/8

N2 - In this technical note, we formulate a fully Bayesian approach for spatio-temporal Gaussian process regression such that multifactorial effects of observations, measurement noise and prior distributions are all correctly incorporated in the predictive distribution. Using discrete prior probabilities and compactly supported kernels, we provide a way to design sequential Bayesian prediction algorithms in which exact predictive distributions can be computed in constant time as the number of observations increases. For a special case, a distributed implementation of sequential Bayesian prediction algorithms has been proposed for mobile sensor networks. An adaptive sampling strategy for mobile sensors, using the maximum a posteriori (MAP) estimation, has been proposed to minimize the prediction error variances. Simulation results illustrate the practical usefulness of the proposed theoretically-correct algorithms.

AB - In this technical note, we formulate a fully Bayesian approach for spatio-temporal Gaussian process regression such that multifactorial effects of observations, measurement noise and prior distributions are all correctly incorporated in the predictive distribution. Using discrete prior probabilities and compactly supported kernels, we provide a way to design sequential Bayesian prediction algorithms in which exact predictive distributions can be computed in constant time as the number of observations increases. For a special case, a distributed implementation of sequential Bayesian prediction algorithms has been proposed for mobile sensor networks. An adaptive sampling strategy for mobile sensors, using the maximum a posteriori (MAP) estimation, has been proposed to minimize the prediction error variances. Simulation results illustrate the practical usefulness of the proposed theoretically-correct algorithms.

UR - http://www.scopus.com/inward/record.url?scp=84864739293&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84864739293&partnerID=8YFLogxK

U2 - 10.1109/TAC.2011.2179430

DO - 10.1109/TAC.2011.2179430

M3 - Article

AN - SCOPUS:84864739293

VL - 57

SP - 2078

EP - 2084

JO - IEEE Transactions on Automatic Control

JF - IEEE Transactions on Automatic Control

SN - 0018-9286

IS - 8

M1 - 6099560

ER -