基于累计时延统计的传感器网络数据同步算法

Accumulative-based delay statistics for data synchronization algorithm wireless sensor network

传感器网络数据同步是数据融合算法正确运行的前提,具体指网内各节点汇报的数据基于同一时间基准,其采集时间、先后顺序等与真实情况一致.然而由于节点晶振的频率偏差和不同的初始计时时刻,网内节点的本地时钟不同步,这使得根据本地时钟标记的数据不能保持同步.提出了一种基于累计时延统计的数据同步算法,通过在数据包头附加一个时延字段,沿途节点根据该数据包的停留时间更新该字段,数据到达远方站点时即包含了数据的总时延,接收站根据当前时刻和累积时延计算数据的采集时间,最终达到数据同步.分析表明该算法可达到ms级同步精度,适合于中低精度应用.相比于常规同步算法,其通信开销几乎为零.

As a premise of correct data fusion in the sensor network, data synchronization means that all the data reported by the sensors must be based on a same time benchmark, the data collected time and its order must be consistent with the real case. Because of the drift of the oscillator and the different initial start-up time, all the local clocks in the network are not synchronized, the data based on the local clock timestamp cannot by synchronized. A new data synchronization algorithm was proposed based on accumulated delay static. A delay field was added in the packet header to record the total undergoing delay since the data were collected by the source node. The internal node would update the field based on the delay. When the sink received the packet, it could induce the collected time of the data with the total delay field and its clock. And all the data could be synchronized at the sink. It is proved that this algorithm can achieve ms level precision and is suited for the medial & low precision level application. Its communication overhead is almost zero compared with the general data synchronization algorithms.