|
Activity Number:
|
421
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 6, 2008 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statisticians in Defense and National Security
|
| Abstract - #300256 |
|
Title:
|
Energy-Performance Issues for Statistical Inference in Large Random Networks
|
|
Author(s):
|
Animashree Anandkumar and Joseph Yukich and Ananthram Swami and Lang Tong*+
|
|
Companies:
|
Cornell University and Lehigh University and Cornell University and Cornell University
|
|
Address:
|
384 Frank H.T. Rhodes Hall, Ithaca, NY, 14850,
|
|
Keywords:
|
distributed detection ; markov random fields ; minimum energy routing ; error exponent ; Stabilizing graph functionals
|
|
Abstract:
|
The issues of energy consumed in data fusion of a multi-hop sensor network and the resulting inference performance at the fusion center are considered. The correlation between the sensor measurements is incorporated via a Markov random field model with a distance-dependent correlation strength. The sensor placement is assumed to be i.i.d. according to a given distribution and its effect on energy and performance is studied. For Poisson and uniformly placed sensors, there is an optimal sensor density that maximizes the error exponent subject to a constraint on the average routing energy. This optimal density crucially depends on the ratio between the measurement variances under the two hypotheses and displays a threshold behavior. Below the threshold, the optimal density tends towards infinity. Above the threshold, it is the minimum feasible value delivering the likelihood ratio.
|
