Wireless Sensor Network Target Tracking Key Technologies

Wireless Sensor Networks (WSN) is a hot research in the cross-disciplinary field of computer, communications, networks, information and control. However, there are still many problems in theory and applications. This thesis focuses on the key technologies about target tracking based on WSN, including the self-organization technology, collaborative information acquirement and multi-target tracking problems. The experimental platform is also constructed. The main contributions are as follows:1. We introduce some self-organization algorithms in target tracking of WSN. A dynamic collaborative self-organization algorithm, named DCS, is presented. In the DCS, the clustering head can be adaptively switched and its member nodes can be added or deleted by online optimization. Moreover, the simulation analysis of the DCS is proposed and the corresponding conclusions are made. It is shown that the DCS can avoid the problem of "too frequent head switches" and save the energy in communications, compared with the classical Information-driven sensor querying (IDSQ) algorithm.2. The optimal tradeoff between the information acquirement and the energy saving is a hot issue in the field of WSN, and hence choosing the information utility function (IUF) is the key problem in above optimization. Here the target-tracking oriented application is considered to exploring the effect of the multiple typical IUFs on the performance of the DCS. Such all-round simulations include parameter design of each IUF, performance analysis (including tracking accuracy and communication hops), computational-complexity analysis, and parameter-robustness analysis. Finally three conclusions about the choice of the IUF are made.3. Different from the traditional target tracking, the data association of the WSN based target tracking faces the novel problem of echo combination, i.e. the signals oriented from different targets may be outputted as a single measurement. Through introducing the additional characteristic information, we propose a simple but effective data association method, which obtains the equivalent measurement, the weighted centroid based on the IUF. Simulation results show that our algorithm remains almost same tracking accuracy, but less computational complexity of the algorithm, and lower loss tracking rate, compared with the Bayesian filtering...