期刊
IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS
卷 15, 期 10, 页码 6638-6650出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TWC.2016.2586844
关键词
Wireless sensor networks; distributed localization; range-based localization algorithms; Bayesian updates; precision agriculture
资金
- Natural Sciences and Engineering Research Council of Canada (Discovery Grants Program)
- SemiosBio, Vancouver, BC, Canada
In this paper, we propose a received signal strength indication-based distributed Bayesian localization algorithm based on message passing to solve the approximate inference problem. The algorithm is designed for precision agriculture applications, such as pest management and pH sensing in large farms, where greater power efficiency besides communication and computational scalability is needed but location accuracy requirements are less demanding. Communication overhead, which is a key limitation of popular non-Bayesian and Bayesian distributed techniques, is avoided by a message passing schedule, in which outgoing message by each node does not depend on the destination node, and therefore is a fixed size. Fast convergence is achieved by: 1) eliminating the setup phase linked with spanning tree construction, which is frequent in belief propagation schemes and 2) the parallel nature of the updates, since no message needs to be exchanged among nodes during each update, which is called the coupled variables phenomenon in non-Bayesian techniques and accounts for a significant amount of communication overhead. These features make the proposed algorithm highly compatible with realistic wireless sensor network (WSN) deployments, e.g., ZigBee, that are based upon the ad hoc ondemand distance vector, where route request and route reply packets are flooded in the network during route discovery phase.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据