Home // International Journal On Advances in Intelligent Systems, volume 4, numbers 3 and 4, 2011 // View article

Optimal State Estimation under Observation Budget Constraints

Authors:
Praveen Bommannavar
Nicholas Bambos

Keywords: monitoring; surveillance; budget; resource allocation; dynamic programming; convexity; optimal estimation

Abstract:
In this paper, we consider the problem of monitoring an intruder in a setting where the number of opportunities to conduct surveillance is budgeted. Specifically, we study a problem in which we model the state of an intruder in our system with a Markov chain of finite state space. These problems are considered in a setting in which we have a hard limit on the number of times we may view the state. We consider the Markov chain together with an associated metric that measures the distance between any two states. We develop a policy to optimally (with respect to the specified metric) keep track of the state of the chain at each time step over a finite horizon when we may only observe the chain a limited number of times. The tradeoff captured is the budget for surveillance versus having a more accurate estimate of the state; the decision at each time step is whether or not to use an opportunity to observe the process. We also examine a scenario in which there is a budget constraint as described as well as a cost on observation. Finally, theoretical properties of the solution are presented. Hence, we present the problem of monitoring the state of an intruder using a Markov chain approach and present an optimal policy for estimating the intruder’s state.

Pages: 57 to 67

Copyright: Copyright (c) to authors, 2011. Used with permission.

Publication date: April 30, 2012

Published in: journal

ISSN: 1942-2679