Multiple Targets Localization And Tracking In Complicated Environments

With the development of modern information technology,the wireless position detection system caused more attention of both academic and industrial communities.The reason is that the wireless position detection system has many outstanding advantages,such as environmental adaptability and the excellent performance of target detection.Affected by the factors of nature surroundings and human,a detection device is urgently demanded to detect targets in the complex environment,by which rich information with complex conditions can be obtained.The wireless position detection system is a bandwidth wireless signal based device,which uses multiple detection nodes to achieve non-intrusive detection for a specific target in different perspective.In this thesis,based on the wireless position detection system,we studied the dense multipath transmission phenomenon,and the inaccurate factors of detecting nodes during deployment.There are mainly two aspects:In this thesis,under the case of the dense multipath phenomenon in complex environment,we proposed a multipath signal matching based,multiple targets and low complexity detection algorithm.Instead of using channel statistical characteristics,this algorithm can achieve multiple targets detection only based on the measured value of Time of Arrival(TOA).The detail principle is to judge whether the TOA of different detection nodes belong to the direct path of the same target by matching,then the GaussNewton method is used on the direct path of the same target to achieve position.This algorithm can not only ensure the accuracy of position,but also reduce the computational complexity.We also studied the algorithm parameters selection and system detection nodes' configuration to achieve best performance.Moreover,in practice,due to the factors of position error and network deployment,the information about the detection nodes' location is always inaccurate.In this thesis,we studied the inaccurate robust positioning and tracking for the detection nodes' location,and proposed a factor graph based parameter belief propagation algorithm.By detecting the information transmitted between the detection node and the target,this algorithm can achieve target position and node detection.Because It is difficult for the likelihood function of message transmission to follow the Gaussian distribution,we employed the First-order Taylor formula to make the likelihood function's nonlinear part be linear.This method can solve the difficult integral and reduce the computational complexity.Moreover,based on the real-time belief propagation algorithm,we tracked the moving target,which can improve the tracking accuracy of the algorithm significantly.In this thesis,based on the above two aspects,our work can solve the key technical difficulties of wireless position detection system in practical application.Simulations were conducted to verify the feasibility of the application,which provide important support and guidance for future system design and production.