Please use this identifier to cite or link to this item:
|Title:||Localized recursive estimation in energy constrained wireless sensor networks||Authors:||Wang, B.
Wireless sensor networks
|Issue Date:||2006||Citation:||Wang, B.,Chua, K.C.,Srinivasan, V. (2006). Localized recursive estimation in energy constrained wireless sensor networks. Journal of Networks 1 (2) : 18-26. ScholarBank@NUS Repository.||Abstract:||This paper proposes a localized recursive estimation scheme for parameter estimation in wireless sensor networks. Given any parameter of a target occurring at some location and time, a number of sensors recursively estimate the parameter by using their local measurements of the parameter that is attenuated with the distance between a sensor and the target location and corrupted by noise. Compared with centralized estimation schemes that transmit all encoded measurements to a sink (or a fusion center), the recursive scheme needs only to transmit the final estimate to a sink. When the sink is faraway from the sensors and multihop communications have to be used, using localized recursive estimation can help to reduce energy consumption and reduce network traffic load. A sensor sequence with the fastest convergence rate is identified, by which the variance of estimation error reduces faster than all other sequences. In the case of adjustable transmission power, a heuristic has been proposed to find a sensor sequence with the minimum total transmission power when performing the recursive estimation. Numerical examples have been used to compare the performance of the proposed scheme with that of a centralized estimation scheme and have also shown the effectiveness of the proposed heuristic. © 2006 ACADEMY PUBLISHER.||Source Title:||Journal of Networks||URI:||http://scholarbank.nus.edu.sg/handle/10635/56514||ISSN:||17962056|
|Appears in Collections:||Staff Publications|
Show full item record
Files in This Item:
There are no files associated with this item.
checked on Apr 19, 2019
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.