| Summary: | 碩士 === 國立清華大學 === 資訊系統與應用研究所 === 101 === Wireless sensor network (WSN) has been applied in difficult environments for long. In a WSN, the base station collects information from a number of sensors, taking measurements such as temperature, humidity and light. Since the sensors are equipped with limited power energy, they work in cycles. That is, the sensors are active in a fixed duration and sleep until next cycle. The frequencies of cycles may vary due to different accuracy or environmental requirements, determining the portion of active time and energy consumption. In order to improve energy efficiency, an existing work [WY11], local synchronization (LS), enables each sensor synchronizing with its neighbors. However, this work deals with local synchronization only in tree topology, not graph topology, and is not robust against sensor failures. To overcome the weakness, we propose a query processing technique for locally synchronized multi-path WSN. Specifically, given a routing graph and the sampling frequencies of sensors, we present a dynamic programming algorithm to find out the optimal network frequencies of sensors such that the WSN is locally synchronized. Furthermore, we demonstrate the experimental results both in real and synthetic datasets, validating the theoretical results and advantages of the proposed approach.
|
|---|
