A Research On The Localization And Tracking Method In Wireless Sensor Network

Wireless sensor network (WSN), which combines the knowledge of sensors, embedded computing, network and wireless communication, is a novel technology on acquiring and processing information. It can be applied to detecting, sensing, collecting and processing the information of monitored objects or environments in the real time and transmit the signal back to whoever needs it. As a new research area in computer science, WSN has a wide application in future, such as in the national defense, environment detection, medical treatment and so on.The background of the research in this dissertation is based on a huge building rescue system founded on wireless sensor networks. The system combines wireless sensor network with the traditional system, and realizes the functions of locating the person, predicting the disaster trend, planning optimal trace and so on. Going into the details of the localization function in the system, this dissertation mainly studies three aspects:self-localization, target location and target tracking.The self-localization is the foundation of localization research in WSN, and a distributed grid location scheme based on Euclidean distance is presented in this dissertation. The main idea of the scheme is to obtain the unknown node location by matching hop count information between itself and the reference nodes. Due to the low localization accuracy and large energy consumption in grid-based location estimation scheme, the proposed scheme uses Euclidean distance as a substitute for matching threshold in order to improve the localization accuracy; moreover, distributed processing is applied to save energy. Simulation results show that the scheme can robustly determine the location of unknown node quickly and accurately, and save a great deal of energy in the whole network.For locating the trapped person in huge building, a distributed EM localization scheme based on acoustic energy is presented in this dissertation. The centralized localization method in WSN needs a great deal of energy consumption, therefore distributed method has been applied to improve this problem. The proposed scheme can process the distributed information from the sensors by using EM algorithm to get the location and energy of the source. Furthermore, loop-based distributed method and cluster-based distributed method are also presented in the dissertation according to the information transfer mode. Simulation illustrates that the localization accuracy of the proposed scheme is maintained, and the scheme can save large amount of energy.How to obtain the track of the trapped person is the key issue in the huge building rescue system. In wireless sensor network, the tracking method based on particle filter has been accepted generally, however a large number of experiments illustrate that particle filter can only capture the moving trajectory with simple changes. In this dissertation, an adaptive variable rate particle filter algorithm is suggested, which is able to adaptively adjust the sampling period by comparing the relationship between particle value and measurement. The method has improved the tracking accuracy of complex moving target. Simulation illustrates that the adaptive variable rate particle filter algorithm has a good performance on complex moving target, especially as the target trajectory has some drastic changes.