中文摘要：

针对在传感器网络目标跟踪的实际应用中,节点感知的数据与目标真实状态之间通常呈现非线性的特点,提出了一种基于改进不敏卡尔曼滤波的传感器网络目标跟踪算法。通过引入粒子群技术对不敏卡尔曼滤波中δ采样点的分布和收敛速度进行优化,使得δ采样点的分布更加接近目标的真实状态,以提高目标跟踪精度。同时,构建了一个随目标移动而动态生成的树形结构作为算法的执行平台。仿真结果表明,采用动态生成树作为算法执行平台提高了节点资源的利用率,降低了网络能耗,采用粒子群优化后的不敏卡尔曼滤波提高了目标跟踪精度,减少了算法运行时间。

英文摘要：

For the nonlinear characteristic between the data sensed by the nodes and the real state of the target in the target tracking application of the sensor networks,a target tracking algorithm based on the improved unscented Kalman filter is presented.The distribution of the δ sampling points in the Unscented Kalman Filter and the convergence speed are optimized by introducing the particle swarm technology,which makes the distribution of the δ sampling points more close to the real state of the targets to improve the accuracy of the target tracking.Meanwhile,a dynamic generated tree following the movement of the targets is constructed to play as the platform to execute the algorithm.Simulation results show that utilizing the dynamically generated tree to play as the algorithm executing platform,the use of the nodes＇ resources can be improved,and the networks energy consumption can be decreased.The unscented Kalman filter optimized by the particle swarm can improve the accuracy of the target tracking and reduce the executing time of the algorithm.