Research On Key Techniques Of Indoor Wireless Sensor Network Tracking

In indoor tracking applications of wireless sensor networks, how to effectively improve the positioning accuracy by combining the low power consumption of wireless sensor networks, the characteristics of the low transmission rate and the complex indoor radio environment has been an urgent problem to solve. The following work had been done mainly on the perspective of reducing the network communication load and handling the tracking and locating problem in complex situation.Firstly, a kind of optimal quantization strategy based on quantized Kalman Filter was proposed. Multi-bit quantized Kalman filter algorithem using the optimal quantizer preserve the useful information as much as possible, while reducing the communication load. At the beginning I explored the rationality of quantifying the observation information utilizing the correlation of the sharing state observed by different nodes in wireless sensor network. Then the shortcomings of the single-bit quantized Kalman filter algorithm were discussed. Finally aiming at the shortcomings of the single-bit quantized Kalman filter, optimal quantized threshold estimator quantization strategy based on multi-bit quantized Kalman filter was discussed. Considering the deficiencies of the single-bit quantized Kalman filter algorithm, the idea of multi-bit quantized Kalman filter algorithm was proposed, and the optimal quantized threshold estimator quantization was designed. A linear motion model simulation results validated the strategy of the best bits multi-bit quantized strategy and the consistency of the optimal quantized Kalman filter algorithm.Secondly, discussing the key technology of the tracking and localizing problem in complex large scale system environment. Using the spatial decomposition algorithem to divide the global system into distributed sub-systems which are appropriate for WSN application. Facing with the problems of the overlapped sub-systems and the shared states, introduce the bipartite fusion graph algorthem to fuse the observation. In the non-linear motion target tracking and locating scene based on the previous study, the linearized approximation of the non-linear motion model of the target was made by ultilizaing the extended Kalman filter algorithm. Then, the previous modified model can be estimated by the optimal quantifyed Kalman filter. The consistency of extended Kalman filter algorithm based on the optimal quantized strategy was verified by the simulation of the nonlinear model. And the results were compared with linear model.