Type-Based Detection with a Fusion Center Performing the Sequential Test in Wireless Sensor Networks

Dmitry KRAMAREV
Insoo KOO
Kiseon KIM

Publication
IEICE TRANSACTIONS on Communications   Vol.E90-B    No.12    pp.3354-3361
Publication Date: 2007/12/01
Online ISSN: 1745-1345
DOI: 10.1093/ietcom/e90-b.12.3354
Print ISSN: 0916-8516
Type of Manuscript: Special Section PAPER (Special Section on Ubiquitous Sensor Networks)
Category: 
Keyword: 
decentralized detection,  sequential test,  method of types,  Gaussian approximation,  wireless sensor networks,  

Full Text: PDF>>
Buy this Article


Summary: 
In this paper, we propose a sequential type-based detection scheme for wireless sensor networks in the case of spatially and temporally identically and independently distributed observations. First, we investigate the optimal sequential detection rule of the proposed scheme, and then with the motivation of reducing the computational complexity of the optimal detection rule, we consider an approximation scheme and derive a suboptimal detection rule. We also compare the performances of the type-based sequential detection scheme with those of the non-sequential type-based detection scheme in terms of both average number of observations and total energy consumption, and determine the region of individual node power where the proposed scheme outperforms the non-sequential scheme. In addition, we show that the approximated detection rule provides the similar results as the optimal detection rule with a significant reduction of the computational complexity, which makes the approximated detection rule useful for real-time applications.