Spatial prediction with mobile sensor networks using Gaussian process regression based on Gaussian Markov random fields

Yunfei Xu, Jongeun Choi

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

Abstract

In this paper, a new class of Gaussian processes is proposed for resource-constrained mobile sensor networks. Such a Gaussian process builds on a GMRF with respect to a proximity graph over a surveillance region. The main advantages of using this class of Gaussian processes over standard Gaussian processes defined by mean and covariance functions are its numerical efficiency and scalability due to its built-in GMRF and its capability of representing a wide range of non-stationary physical processes. The formulas for Bayesian posterior predictive statistics such as prediction mean and variance are derived and a sequential field prediction algorithm is provided for sequentially sampled observations. For a special case using compactly supported kernels, we propose a distributed algorithm to implement field prediction by correctly fusing all observations in Bayesian statistics. Simulation results illustrate the effectiveness of our approach.

Original languageEnglish
Title of host publicationASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011
Pages173-180
Number of pages8
DOIs
Publication statusPublished - 2011 Dec 1
EventASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011 - Arlington, VA, United States
Duration: 2011 Oct 312011 Nov 2

Publication series

NameASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011
Volume2

Other

OtherASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011
CountryUnited States
CityArlington, VA
Period11/10/3111/11/2

Fingerprint

Sensor networks
Wireless networks
Statistics
Parallel algorithms
Scalability

All Science Journal Classification (ASJC) codes

  • Fluid Flow and Transfer Processes
  • Control and Systems Engineering

Cite this

Xu, Y., & Choi, J. (2011). Spatial prediction with mobile sensor networks using Gaussian process regression based on Gaussian Markov random fields. In ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011 (pp. 173-180). (ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011; Vol. 2). https://doi.org/10.1115/DSCC2011-6092
Xu, Yunfei ; Choi, Jongeun. / Spatial prediction with mobile sensor networks using Gaussian process regression based on Gaussian Markov random fields. ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011. 2011. pp. 173-180 (ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011).
@inproceedings{cb83f438699449c9bb10d681f0aec28e,
title = "Spatial prediction with mobile sensor networks using Gaussian process regression based on Gaussian Markov random fields",
abstract = "In this paper, a new class of Gaussian processes is proposed for resource-constrained mobile sensor networks. Such a Gaussian process builds on a GMRF with respect to a proximity graph over a surveillance region. The main advantages of using this class of Gaussian processes over standard Gaussian processes defined by mean and covariance functions are its numerical efficiency and scalability due to its built-in GMRF and its capability of representing a wide range of non-stationary physical processes. The formulas for Bayesian posterior predictive statistics such as prediction mean and variance are derived and a sequential field prediction algorithm is provided for sequentially sampled observations. For a special case using compactly supported kernels, we propose a distributed algorithm to implement field prediction by correctly fusing all observations in Bayesian statistics. Simulation results illustrate the effectiveness of our approach.",
author = "Yunfei Xu and Jongeun Choi",
year = "2011",
month = "12",
day = "1",
doi = "10.1115/DSCC2011-6092",
language = "English",
isbn = "9780791854761",
series = "ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011",
pages = "173--180",
booktitle = "ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011",

}

Xu, Y & Choi, J 2011, Spatial prediction with mobile sensor networks using Gaussian process regression based on Gaussian Markov random fields. in ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011. ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011, vol. 2, pp. 173-180, ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011, Arlington, VA, United States, 11/10/31. https://doi.org/10.1115/DSCC2011-6092

Spatial prediction with mobile sensor networks using Gaussian process regression based on Gaussian Markov random fields. / Xu, Yunfei; Choi, Jongeun.

ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011. 2011. p. 173-180 (ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011; Vol. 2).

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

TY - GEN

T1 - Spatial prediction with mobile sensor networks using Gaussian process regression based on Gaussian Markov random fields

AU - Xu, Yunfei

AU - Choi, Jongeun

PY - 2011/12/1

Y1 - 2011/12/1

N2 - In this paper, a new class of Gaussian processes is proposed for resource-constrained mobile sensor networks. Such a Gaussian process builds on a GMRF with respect to a proximity graph over a surveillance region. The main advantages of using this class of Gaussian processes over standard Gaussian processes defined by mean and covariance functions are its numerical efficiency and scalability due to its built-in GMRF and its capability of representing a wide range of non-stationary physical processes. The formulas for Bayesian posterior predictive statistics such as prediction mean and variance are derived and a sequential field prediction algorithm is provided for sequentially sampled observations. For a special case using compactly supported kernels, we propose a distributed algorithm to implement field prediction by correctly fusing all observations in Bayesian statistics. Simulation results illustrate the effectiveness of our approach.

AB - In this paper, a new class of Gaussian processes is proposed for resource-constrained mobile sensor networks. Such a Gaussian process builds on a GMRF with respect to a proximity graph over a surveillance region. The main advantages of using this class of Gaussian processes over standard Gaussian processes defined by mean and covariance functions are its numerical efficiency and scalability due to its built-in GMRF and its capability of representing a wide range of non-stationary physical processes. The formulas for Bayesian posterior predictive statistics such as prediction mean and variance are derived and a sequential field prediction algorithm is provided for sequentially sampled observations. For a special case using compactly supported kernels, we propose a distributed algorithm to implement field prediction by correctly fusing all observations in Bayesian statistics. Simulation results illustrate the effectiveness of our approach.

UR - http://www.scopus.com/inward/record.url?scp=84881473115&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84881473115&partnerID=8YFLogxK

U2 - 10.1115/DSCC2011-6092

DO - 10.1115/DSCC2011-6092

M3 - Conference contribution

SN - 9780791854761

T3 - ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011

SP - 173

EP - 180

BT - ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011

ER -

Xu Y, Choi J. Spatial prediction with mobile sensor networks using Gaussian process regression based on Gaussian Markov random fields. In ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011. 2011. p. 173-180. (ASME 2011 Dynamic Systems and Control Conference and Bath/ASME Symposium on Fluid Power and Motion Control, DSCC 2011). https://doi.org/10.1115/DSCC2011-6092