【作者】 匡兴红；

【导师】 邵惠鹤；

【作者基本信息】 上海交通大学 ， 控制理论与控制工程， 2008， 博士

【摘要】 无线传感器网络集成了传感器技术、微机电技术、分布式信息处理技术和无线通信技术,它已成为当前IT领域研究的热点之一。由于网络自组织、广覆盖、高容错性以及高精度测量等固有优点以及组网成本低、构建灵活、方便等特点,使得无线传感器网络在军事、民用等领域应用广泛。本文深入研究了无线传感器网络中的定位跟踪问题,包括节点定位、生化气体源定位以及目标跟踪问题,并设计开发了面向定位跟踪应用研究的平台,为进一步的应用研究打下基础。在上述问题的研究中,针对不同的实际应用场景提出了具有理论和实际价值的定位、跟踪新方法,并通过计算机仿真对性能进行了验证和评价。具体地,论文主要的研究成果和创新有:(1)无线传感器网络中的节点定位研究将无线传感器网络中的节点定位问题看作目标跟踪的对偶问题,提出一种基于虚拟信标点的节点定位方案,区别于基于固定信标点方式的现有定位算法。基于超球面交叉机制,提出了VB-ERL节点定位算法。它利用非线性最小二乘优化方法,求解由能量比对构成的最小化成本函数实现节点自我定位。为了克服VB-ERL节点定位算法存在的弊端,对节点定位问题又提出了一种贝叶斯框架下基于改进粒子滤波的VB-IPF算法,并实现在线迭代预估。(2)无线传感器网络中的生化气体源定位研究将无线传感器网络应用于生化气体源定位,以提高人们对突发事件的快速反应能力。首先提出了一种鲁棒性强的极大似然定位算法,对比研究了直接三边法以及非线性最小二乘法。为了消除集中式定位算法的固有缺陷及节省网络能量,将分层传感器网络应用于生化源定位,提出了一种分层传感器网络下的基于改进粒子滤波的分布式定位算法,通过子区域管理节点的状态传递,循环迭代预估得到收敛的生化气体源位置。(3)无线传感器网络中的目标定位跟踪研究目标定位跟踪技术无论是在军事还是民用领域都有着重要的应用价值。首先研究了传感器网络下一种仅有角度测量的移动目标定位跟踪。从提高滤波算法的角度出发来提高目标预估定位精度,提出了一种无线传感器网络下的集中式融合方案,应用SRCDKF-PF算法及其改进措施进行预估。其次,考虑无线传感器网络中的目标跟踪节能的需要,提出了一种轻量级的自适应机制的分布式目标跟踪算法。在跟踪算法中:采用节点密度自适应的节点激活机制,寻找定位精度与网络能耗之间的平衡;提出的退避定时机制用于头节点的选取及传输序列的确定;基于改进粒子滤波的定位算法提高了滤波的精度;提出的一种波浪式区域逐层唤醒机制用于快速重新定位丢失的目标。(4)面向定位跟踪的平台设计与开发面向定位跟踪的平台开发,包括无线传感器网络网关的开发、传感器节点的开发。开发了基于GPRS网络通讯的网络网关,实现与Internet外部网的连接。开发了基于CC1010传感器节点,重点阐述了RF通讯的设计。为进一步的应用和研究打下基础。更多还原

【Abstract】 Wireless Sensor Network (WSN), which integrates the sensor, micro-electro-mechanism system (MEMS), wireless communication and distributed information processing, becomes a hot research topics at the current Information Technology field. The WSN has many advantages such as self-organization, high accuracy measurements, and high fault-toleration, widely coverage et al. Characteristics of low cost, low power, flexible deployment make it more widely applied ranging from the military areas to the abundant civil areas.The dissertation deeply studied the localization and target tracking problems in wireless sensor network, including node self-localization, plume source localization and the target tracking problems. General localization and target tracking oriented platform is developed, which serves for further studying and practical applications. Due to the problems listed above, many new localization and tracking algorithms are proposed according to the practical applications in WSN, these algorithms have a certain values on theories and practical applications. Extensive simulations have been conducted to validate and evaluate the performance of the proposed algorithms.The concrete contents of this dissertation are given as below:(1) Research on node localization in wireless sensor networkNode localization in wireless sensor networks is treated as a functional dual of target tracking from a novel perspective in the dissertation. A node localization scheme is proposed based on the virtual beacons, which is different from the current fixed beacons node localization algorithm. A VB-ERL localization algorithm is proposed based on the hyper-sphere intersection mechanism. The nodes estimate their locations by solving the least cost function generated by many pairs of received virtual beacons-energy ratios. In order to eliminate the drawbacks of the VB-ERL algorithm, a localization algorithm named VB-IPF which based on the improved particle filter under Bayesian Frame is also proposed. It estimates the node location on line.(2) Research on plume source localization in wireless sensor networkA wireless sensor networks is proposed to estimate the plume source location. Such WSN can be of tremendous help to emergency personnel trying to protect people from terrorist attacks or responding to an accident. First, a robust centralized localization algorithm named maximum likelihood estimation (MLE) algorithm is proposed. It’s performance is evaluated compared with the direct Trilateration (DT) algorithm and the nonlinear least square (NLS) algorithm. In order to save the energy and eliminate the various problems associated with the centralization algorithms, a hierarchical wireless sensor network is proposed to estimate the plume source location. The performed distributed algorithm is based on the improved particle filter. The source location is obtained by the manager nodes’estimation state transmission in each sub-region in turn.(3) Research on target tracking in wireless sensor networkTarget tracking applications in wireless sensor network have a great value ranging from the military to the abundant civil areas.A bearings-only tracking for maneuvering target was firstly studied in wireless sensor network. A centralized fusion scheme in WSN was proposed to estimate the target based on the SRCDKF-PF algorithm and its improved methods. It improves the target estimation accuracy from the aspect of the filters improvements.Second, a new adaptive, distributed lightweight target tracking algorithm is proposed considering the energy saving requirement in WSN. In the algorithm, an adaptive mechanism based on the node density is proposed to obtain optimal active region size. A backoff timer mechanism is proposed to elect the leader node and obtain the transmission order of the localization nodes’measurements. A localization mechanism based on the improved particle filter improves the tracking performance. A step by step recovery mechanism is proposed to relocate the missing target.(4)Research on target tracking oriented platform of WSNA target tracking oriented platform of WSN is designed and developed, including gateway and sensor node. The gateway implements the reliable link between WSN and the other networks such as Internet by TCP/IP protocol through the GPRS communication system. The sensor node based on CC1010 chip implements the communication among nodes in WSN. It offers a general target tracking oriented platform for studying of the future related problems in WSN.更多还原