A direct algorithm for joint optimal sensor scheduling and MAP state estimation for hidden Markov models

David Jun, David M. Cohen, Douglas L Jones

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

Abstract

Sensing systems with multiple sensors and operating modes warrant active management techniques to balance estimation quality and measurement costs. Existing literature shows that in the joint sensor-scheduling and state-estimation problem for HMMs, estimator optimization can be done independently of the scheduler at each time step. We investigate the special case when a MAP estimator is used, and show how the joint problem can be converted to a standard Partially Observable MarkovDecision Process (POMDP), which in turn enables us to use POMDP solvers. As this approach is highly redundant, we derive a direct solution, which exploits the separability property while still utilizing standard solvers. When compared to standard techniques, the direct algorithm provides savings by a factor of the state-space dimension. Numerical results are given for an example motivated by wildlife monitoring.

Original languageEnglish (US)
Title of host publication2013 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 - Proceedings
Pages4212-4215
Number of pages4
DOIs
StatePublished - Oct 18 2013
Event2013 38th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 - Vancouver, BC, Canada
Duration: May 26 2013May 31 2013

Other

Other2013 38th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013
CountryCanada
CityVancouver, BC
Period5/26/135/31/13

Fingerprint

State estimation
Hidden Markov models
Scheduling
Sensors
Monitoring
Costs

Keywords

  • controlled HMM
  • POMDP
  • sensor management

ASJC Scopus subject areas

  • Software
  • Signal Processing
  • Electrical and Electronic Engineering

Cite this

Jun, D., Cohen, D. M., & Jones, D. L. (2013). A direct algorithm for joint optimal sensor scheduling and MAP state estimation for hidden Markov models. In 2013 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 - Proceedings (pp. 4212-4215). [6638453] https://doi.org/10.1109/ICASSP.2013.6638453

A direct algorithm for joint optimal sensor scheduling and MAP state estimation for hidden Markov models. / Jun, David; Cohen, David M.; Jones, Douglas L.

2013 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 - Proceedings. 2013. p. 4212-4215 6638453.

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

Jun, D, Cohen, DM & Jones, DL 2013, A direct algorithm for joint optimal sensor scheduling and MAP state estimation for hidden Markov models. in 2013 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 - Proceedings., 6638453, pp. 4212-4215, 2013 38th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013, Vancouver, BC, Canada, 5/26/13. https://doi.org/10.1109/ICASSP.2013.6638453
Jun D, Cohen DM, Jones DL. A direct algorithm for joint optimal sensor scheduling and MAP state estimation for hidden Markov models. In 2013 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 - Proceedings. 2013. p. 4212-4215. 6638453 https://doi.org/10.1109/ICASSP.2013.6638453
Jun, David ; Cohen, David M. ; Jones, Douglas L. / A direct algorithm for joint optimal sensor scheduling and MAP state estimation for hidden Markov models. 2013 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 - Proceedings. 2013. pp. 4212-4215
@inproceedings{a1e4386c8a07420ba5c408e145ad4eb0,
title = "A direct algorithm for joint optimal sensor scheduling and MAP state estimation for hidden Markov models",
abstract = "Sensing systems with multiple sensors and operating modes warrant active management techniques to balance estimation quality and measurement costs. Existing literature shows that in the joint sensor-scheduling and state-estimation problem for HMMs, estimator optimization can be done independently of the scheduler at each time step. We investigate the special case when a MAP estimator is used, and show how the joint problem can be converted to a standard Partially Observable MarkovDecision Process (POMDP), which in turn enables us to use POMDP solvers. As this approach is highly redundant, we derive a direct solution, which exploits the separability property while still utilizing standard solvers. When compared to standard techniques, the direct algorithm provides savings by a factor of the state-space dimension. Numerical results are given for an example motivated by wildlife monitoring.",
keywords = "controlled HMM, POMDP, sensor management",
author = "David Jun and Cohen, {David M.} and Jones, {Douglas L}",
year = "2013",
month = "10",
day = "18",
doi = "10.1109/ICASSP.2013.6638453",
language = "English (US)",
isbn = "9781479903566",
pages = "4212--4215",
booktitle = "2013 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 - Proceedings",

}

TY - GEN

T1 - A direct algorithm for joint optimal sensor scheduling and MAP state estimation for hidden Markov models

AU - Jun, David

AU - Cohen, David M.

AU - Jones, Douglas L

PY - 2013/10/18

Y1 - 2013/10/18

N2 - Sensing systems with multiple sensors and operating modes warrant active management techniques to balance estimation quality and measurement costs. Existing literature shows that in the joint sensor-scheduling and state-estimation problem for HMMs, estimator optimization can be done independently of the scheduler at each time step. We investigate the special case when a MAP estimator is used, and show how the joint problem can be converted to a standard Partially Observable MarkovDecision Process (POMDP), which in turn enables us to use POMDP solvers. As this approach is highly redundant, we derive a direct solution, which exploits the separability property while still utilizing standard solvers. When compared to standard techniques, the direct algorithm provides savings by a factor of the state-space dimension. Numerical results are given for an example motivated by wildlife monitoring.

AB - Sensing systems with multiple sensors and operating modes warrant active management techniques to balance estimation quality and measurement costs. Existing literature shows that in the joint sensor-scheduling and state-estimation problem for HMMs, estimator optimization can be done independently of the scheduler at each time step. We investigate the special case when a MAP estimator is used, and show how the joint problem can be converted to a standard Partially Observable MarkovDecision Process (POMDP), which in turn enables us to use POMDP solvers. As this approach is highly redundant, we derive a direct solution, which exploits the separability property while still utilizing standard solvers. When compared to standard techniques, the direct algorithm provides savings by a factor of the state-space dimension. Numerical results are given for an example motivated by wildlife monitoring.

KW - controlled HMM

KW - POMDP

KW - sensor management

UR - http://www.scopus.com/inward/record.url?scp=84890529618&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84890529618&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2013.6638453

DO - 10.1109/ICASSP.2013.6638453

M3 - Conference contribution

AN - SCOPUS:84890529618

SN - 9781479903566

SP - 4212

EP - 4215

BT - 2013 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2013 - Proceedings

ER -