Optimal Sensor Selection in Binary Heterogeneous Sensor Networks Articles
Overview
published in
publication date
- April 2009
start page
- 1577
end page
- 1587
issue
- 4
volume
- 57
Digital Object Identifier (DOI)
International Standard Serial Number (ISSN)
- 1053-587X
Electronic International Standard Serial Number (EISSN)
- 1941-0476
abstract
- We consider the problem of sensor selection in a heterogeneous sensor network when several types of binary sensors with different discrimination performance and costs are available. We want to analyze what is the optimal proportion of sensors of each class in a target detection problem when a total cost constraint is specified. We obtain the conditional distributions of the observations at the fusion center given the hypotheses, necessary to perform an optimal hypothesis test in this heterogeneous scenario. We characterize the performance of the tests by means of the symmetric Kullback-Leibler divergence, or J -divergence, applied to the conditional distributions under each hypothesis. By formulating the sensor selection as a constrained maximization problem, and showing the linearity of the J-divergence with the number of sensors of each class, we found that the optimal proportion of sensors is ldquowinner takes allrdquo like. The sensor class with the best performance/cost ratio is selected.