Near-Optimal Co-Deployment Of Chargers And Sink Stations In Rechargeable Sensor Networks
Songyuan Li,Lingkun Fu,Shibo He,Youxian Sun
DOI: https://doi.org/10.1145/3070721
2018-01-01
ACM Transactions on Embedded Computing Systems
Abstract:Wireless charging technology has drawn great attention of both academia and industry in recent years, due to its potential of significantly improving the system performance of sensor networks. The emergence of an open-source experimental platform for wireless rechargeable sensor networks, Powercast, has made the theoretical research closer to reality. This pioneering platform is able to recharge sensor nodes much more efficiently and allows different communication protocols to be implemented upon users' demands. Different from the RFID-based model widely used in the existing works, Powercast designs the charger and sink station separately. This leads to a new design challenge of cooperatively deploying minimum number of chargers and sink stations in wireless rechargeable sensor networks. Such a co-deployment issue is extremely challenging, since the deployments of chargers and sink stations are coupled, and each subproblem is known to be NP-hard. The key to the design is to understand the intrinsic relationship between data flow and energy flow, which is interdependent. In this article, we tackle this challenge by dividing it into two subproblems and optimizing charger and sink station deployment iteratively. Specifically, we first transform each subproblem to a max-flow problem. With this, we are able to select chargers or sink stations according to their contributions to the total flow rate. We design greedy-based algorithms with a guaranteed worst-case bound ln R/xi. for the subproblems of charger deployment and sink station deployment, respectively. Further, we address the original problem by designing an iterative algorithm that solves two subproblems alternatively to achieve a near optimal performance. We corroborate our analysis by extensive simulations under practical coefficient settings and demonstrate the advantage of the proposed algorithm.