基于流预测的无线传感器网络动态功率管理

Dynamic power management of wireless sensor networks using stream forecast

为了最大限度地节约能量的使用,延长无线传感器网络使用寿命,提出了一种利用小波和自回归的动态功率管理(DPM)方法.该方法利用收发器(sink)节点上的历史数据流预测未来值,在后续周期内,若传感器节点的观测值不超过给定阈值则不向sink节点发送数据,sink节点将预测值作为观测结果,通过减少传感器节点工作时间,降低网络数据传输量来减少传感器网络的总体能量消耗.理论分析和试验结果表明,该方法是有效的.

An effective method of dynamic power management based on wavelet and AR （auto regression） is proposed to extend life of wireless sensor networks by making economical use of energy. The data stream gathered by sink is used to forecast in this method, and in some future periods nodes do not send back data if their observed values are not out of the threshold, the forecasted values are accepted as the result, so as to reduce the energy consumption of the sensor networks in aspects of less working time and less transported messages. Theory analyses and experiment result show that it is effective.