Files in this item



application/pdfECE499-Fa2012-deng.pdf (785kB)Restricted to U of Illinois
(no description provided)PDF


Title:Adaptive Sampling for Estimation
Author(s):Deng, Mo
Contributor(s):Veeravalli, Venugopal V.
sensor networks
distributed networks
fusion center
adaptive sampling
Abstract:In sensor networks, a large number of independent sensors are located in a distributed manner. Each sensor will send its local statistics to the Fusion Center (FC). The fusion center is in charge of determining which hypothesis is true in a detection problem context and the final parameter estimate in an estimation problem context. Instead of sending local observations directly in the centralized framework, the decentralized framework requires the local statistics sent by each sensor usually being a compressed version (i.e. a non-trivial function) of local observations of that sensor, thus avoiding problems such as congestion in the networks. Traditional decentralized detection and/or parameter estimation problems in sensor networks usually assume that all local sensors sense and send statistics synchronously to the fusion center, with communication rate pre-assumed. In our work, we consider the problem where asynchronous communications between sensors and fusion center are used. Specifically, each sensor only sends statistics when some local decision criterions are satisfied. The fusion center will make global estimates based on the asynchronously updated local statistics under the global rules. The goal of this project is to find the optimal set of local, global and stopping rules in the sense that there is the best trade-off between the performance loss (in terms of the increase of MSE of the decentralized estimates compared to that in the centralized case) and the energy consumption.
Issue Date:2012-12
Publication Status:unpublished
Peer Reviewed:not peer reviewed
Date Available in IDEALS:2014-01-09

This item appears in the following Collection(s)

Item Statistics