无线传感器网络目标跟踪中的节点选择算法研究

发布时间：2018-03-19 02:12

  本文选题：无线传感器网络　切入点：目标跟踪　出处：《南京邮电大学》2017年硕士论文　论文类型：学位论文

【摘要】：无线传感器网络是由随机部署在监控区域内的大量廉价微型传感器节点,通过无线通信方式组成的多跳自组织网络系统,它可以实时采集、处理和传输网络覆盖区域内被感知对象的信息,并把这些信息发送给用户。WSN被广泛应用于军事、智能交通、环境监控和医疗护理等多个领域,其中目标跟踪是无线传感器网络最具代表性的应用之一。由于传感器节点稠密分布,如果将检测到目标的所有传感器节点都用于目标跟踪,虽然跟踪精度很高,但是能量消耗降低了网络生存周期。传感器节点选择算法就是从候选传感器集合中选择节点子集,在满足目标跟踪性能的同时尽可能降低网络能量消耗,延长网络生存周期。针对不同目标跟踪环境,提出几种节点选择算法。主要工作和贡献如下:1.对机动性不强的运动目标,提出基于扩展Kalman滤波的多步预测的节点选择算法。以多步状态预测误差协方差矩阵行列式的加权为目标函数,从候选传感器节点集合选择一组节点最大化该目标函数。在均方根误差和平均剩余能量方面,通过仿真比较该算法与基于扩展Kalman滤波一步预测的节点选择算法。2.针对目标机动特性和传感器测量噪声统计特性未知情况下的目标跟踪问题,利用扩展H∞滤波算法与扩展Kalman滤波算法的形式相类似,提出扩展H∞滤波的类Cramer-Rao下界的节点选择算法。并在跟踪性能和能量消耗方面将该算法与随机节点选择算法和基于目标预测位置最近邻的节点选择方法进行仿真比较。3.在集中式无线传感器网络目标跟踪中,簇头节点在跟踪过程中消耗大量通信和计算能量,容易造成头节点的失效问题,基于扩展Kalman滤波提出顺序处理的目标跟踪算法,将数据处理和通信分配到跟踪簇集内各个传感器上,不需要头节点接收和集中处理其他成员传感器节点的测量数据。在算法计算时间和跟踪性能两方面比较算法优劣。总之,本文围绕无线传感器网络目标跟踪中的节点选择问题展开研究,所得结果不仅具有重要的理论价值,而且具有广泛的实际应用价值。
[Abstract]:Wireless sensor network (WSN) is a multi-hop ad hoc network system, which is composed of a large number of cheap sensor nodes deployed randomly in the monitoring area, and can be collected in real time. It processes and transmits the information of perceived objects in the area covered by the network, and sends the information to the user. WSN is widely used in many fields, such as military, intelligent transportation, environmental monitoring and medical care, etc. Target tracking is one of the most representative applications in wireless sensor networks. Because of the dense distribution of sensor nodes, if all sensor nodes detected the target are used for target tracking, although the tracking accuracy is very high, However, the energy consumption reduces the lifetime of the network. The sensor node selection algorithm is to select the node subset from the candidate sensor set, which can not only meet the target tracking performance, but also reduce the network energy consumption as much as possible. Several node selection algorithms are proposed for different target tracking environments. The main work and contributions are as follows: 1. A node selection algorithm for multistep prediction based on extended Kalman filter is proposed. The objective function is the weighting of the determinant of multistep state prediction error covariance matrix. Select a set of nodes from the set of candidate sensor nodes to maximize the objective function. The algorithm is compared with the node selection algorithm based on one-step prediction of extended Kalman filter by simulation. 2. Aiming at the target tracking problem when the target maneuvering characteristics and the statistical characteristics of sensor measurement noise are unknown, The form of extended H 鈭，

本文编号：1632352