Previous [ 1] [ 2] [ 3] [ 4] [ 5] [ 6] [ 7] [ 8] [ 9] [ 10] [ 11] [ 12] [ 13] [ 14]

Journal of Information Science and Engineering, Vol. 32 No. 2, pp. 439-454 (March 2016)

Power-Aware Classifier Selection in Wireless Sensor Networks*

1Institute of Computing Technology
Chinese Academy of Sciences
Beijing, 100190 P.R. China
2University of Chinese Academy of Sciences
Beijing, 100190 P.R. China
3State Grid Information and Communication Company of Hunan Electric Power Company
Changsha 410007 P.R. China
E-mail: {xiaokejiang; wangxiaofeng; wangrui; lcui}

Many wireless sensor networks (WSNs) based sensing system are as intelligent as possible to accurately sense and recognize interested targets, while large amounts of sensing data generated by WSNs and the limited energy pose great challenges to target classification in WSNs. Thus, it is necessary to obtain tradeoff between power consumption and classification accuracy. Inspired by AdaBoost algorithm, we present a power- aware classification scheme named CSBoost to cluster right classifiers forming strong final classifier. CSBoost scheme can minimize systems power consumption subject to a lower bound on classification accuracy. Specifically speaking, in order to select appropri- ate classifiers, we first give the cost function and utility function according to upper error bound of the final classifier. Then we map the classifier selection problem into 0-1 integer programming problem and provide a heuristic greedy algorithm based method to solve the problem in a polynomial-time. Finally, we conduct experiment on real data to validate and evaluate our proposed scheme. The experimental results demonstrate that CSBoost scheme can get a better performance, comparing with traditional methods.

Keywords: wireless sensor networks, classification accuracy, classifier selection, poweraware classification, AdaBoost algorithm, greedy algorithm, 0-1 integer programming

Full Text () Retrieve PDF document (201603_11.pdf)

Received September 22, 2014; revised November 28, 2014; accepted January 8, 2015.
Communicated by Xiaohong Jiang.
* This paper is supported by National Natural Science Foundation of China (NSFC) under Grant No. 61379134 and 61202211.