摘要
针对集中目标跟踪和分层目标跟踪中心节点通信瓶颈以及容错性能差的不足,提出了一种分布式动态一致性非线性目标跟踪策略。目标状态初始化由网络节点采用加权最小二乘法完成。整个跟踪过程采用动态成簇策略,分阶段选择并唤醒任务节点检测目标并执行分布式一致性扩展卡尔曼滤波策略完成目标的状态估计,其余节点进入休眠状态从而能降低系统的能耗。从跟踪误差和能量两个方面,与集中目标跟踪算法相比,仿真结果表明所提算法与集中卡尔曼滤波相比,跟踪精度相当,适用于要求高可靠度的非线性跟踪。此外分布式的工作方式使得节点仅需与邻居交换数据并在局部完成状态估计,消除集中式结构中心节点的瓶颈,以保证部分传感器节点的损坏不会影响到全局任务的完成。
To overcome shortcoming of center node communication bottleneck constraints and fault tolerance in centralized tar- get tracking and hierarchical object tracking, the paper presented a distributed dynamical consensus nonlinear target tracking strategy. The target state was initialized by the weighted least squares method. The entire tracking process implement dynami- cal clustering strategy, the tasking nodes were selected dynamically and woke up for detection of moving target. Then tasking sensors implement distributed extended Kalman filtering strategy for state estimates, the rest nodes turned into sleep in order to reduce the energy consumption of the system. Compared with central target tracking algorithm from two aspects through simula- tion, i.e. tracking error and energy consumption, the results show that the tracking accuracy is comparable. It is suitable for high reliability nonlinear tracking. In addition, state estimates is completed in distributed manner that nodes only need to ex- change data with their neighbors, it can eliminate the bottleneck of the central node in central structure, in order to ensure that the damage of part sensor nodes will not affect the comoletion of the global task.
出处
《计算机应用研究》
CSCD
北大核心
2013年第5期1365-1369,共5页
Application Research of Computers
基金
国家自然科学基金资助项目(61103108)
国家"863"计划资助项目(2007AA022008)
2012年国家科技支撑计划资助项目(2012BAH08B00
2012BAH08B01)
湖南省教育厅资助项目(12C1123)
