摘要
现有的各种目标跟踪技术未能综合考虑不同目标的运动特征,提出了一种新的基于预测的目标跟踪技术,以减少监控节点数目。根据目标运动的当前测量数据或者历史记录确定目标的运动特征,然后结合目标的当前位置、速度、运动方向等信息预测目标的未来位置;当目标位置预测失败时,网络根据目标的运动历史记录和先验知识逐级启动预测失败恢复过程。仿真结果显示在给定节点与基站分布、节点感知范围和目标运动特性等参数的前提下,比PES方法的目标丢失率大大降低,网络寿命有较大增加,表明采用在确保网络可靠跟踪目标的前提下,减少了被唤醒传感器节点的数目,从而降低了节点的能耗,延长了目标跟踪传感器网络的寿命。
Movement features have not been considered synthetically in all kinds of present object tracking techniques. A new object tracking techniques based on prediction aiming at reducing the number of monitoring nodes is proposed, which at first decides movement features of the object depending on current measurements or history records of its movement, then predicts its future location by integrating with its current information such as location, velocity, and movement direction. When prediction on the object location fails, the network will start hierarchically recovery process from prediction failure according to its records on movement histoy or a priori knowledge. Simulation results show the missing rate for this method is much lower than PES's, and its lifetime increases largely given parameters such as the distribution of the nodes and the base; station, the sense scope of the nodes and the motion characteristic of the objects, which demonstrate the approach reduces the number of sensor nodes waken up, thus lower energy consumption of nodes so as to prolong the longevity of object tracking sensor netowrks.
出处
《计算机仿真》
CSCD
2008年第8期118-122,共5页
Computer Simulation
基金
浙江省自然科学基金资助项目(Y105346)
关键词
无线传感器网络
目标跟踪
运动预测
wireless sensor networks
object tracking
movement prediction