Energy-efficient On-demand Active Contour Service for Sensor Networks

Yangfan Zhou,Junjie Xiong,Michael R. Lyu,Jiangchuan Liu,Kam-Wing Ng

DOI: https://doi.org/10.1109/mobhoc.2009.5336974

2009-01-01

Abstract:Contour mapping is an important technique for Wireless Sensor Networks (WSNs) in environmental monitoring to abstract the information of a monitored field. State-of-the-art approaches for contour mapping, however, are neither energy-optimized, nor capable of handling heterogeneous user requests. In this paper, we develop a novel energy-efficient On-demand Active Contour Service (OACS) for power-constrained WSNs. OACS regresses the field intensity function with kernel Support Vector Regression (SVR), a novel machine learning tool that flexibly handles both contour line and contour map requests. OACS also adaptively accommodates a wide range of contour line/map precision requirements: (1) For applications of low precision, only a minimum set of nodes are scheduled in working mode while others are sleeping for conserving energy. (2) For applications of high precision, through an active and progressive learning algorithm, OACS determines the best set of nodes that should be turned on for improving the contour line/map precision. Evaluation based on diverse realistic models demonstrates that OACS provides quality and seamless contour services for various application requirements yet significantly conserves energy.