Tracking And Localization Method With Wireless Sensor Networks Under Dynamic Propagation Error

With the rapid development of radio positioning and sensor technologies,target tracking and localization has become a challenging task.In real scenarios,due to the interference of various factors such as the limitation of the sensors themselves,the influence of random noise in the surrounding environment and the instability of network transmission error,the measurement data obtained during target tracking process often contains dynamically changing noise,which brings dynamic propagation error and makes information processing more complicated.In addition,the complex noise statistics of the system may sometimes be completely unknown due to the constant changes in time and environmental conditions,which has a very critical impact on the accurate estimation of the target state in the system.For these reasons,aiming at the complex noise environment,this thesis proposes several tracking and localization methods with wireless sensor networks under dynamic propagation error based on existing theoretical foundations.The main research contributions of this thesis are as follows:(1)Aiming at the problem of single sensor node target tracking under the dynamic change of Gaussian noise,this thesis proposes a target tracking and localization algorithm based on unscented Kalman filtering and multi-error model(MEM-UKF).The algorithm takes into account the dynamic change of target observation noise error caused by time-varying propagation error in complex environments,and constructs a multi-error model of target observation based on Gaussian noise distribution.Then the tracking trajectory of the target is obtained by fusing the results of multiple parallel unscented Kalman filters corresponding to the multi-error model.Compared with the traditional algorithm,this algorithm improves the target tracking accuracy under dynamic observation noise error and is more suitable for complex and variable real tracking scenarios.(2)Aiming at the target fusion tracking problem based on wireless sensor networks under the dynamic change of Gaussian noise,this thesis proposes a wireless sensor networks interactive fusion target tracking and localization algorithm based on time-varying observation error(IF-TVOE).Firstly,the algorithm establishes a time-varying error model for the change of observation noise of each sensor node in the wireless sensor networks at different times and environments.Secondly,the joint observations(distance observation,angle observation and Doppler observation)and the proposed time-varying error model are used to model the target state to obtain the prior information of the target state.Finally,the covariance interaction algorithm is used to fuse the results of multiple parallel unscented Kalman filters corresponding to multiple sensor nodes under the dynamic error model to obtain the tracking trajectory of the target,which ensures the consistency of the fusion results.The effectiveness of the proposed algorithm in solving the target tracking problem under dynamic observation noise errors is demonstrated by the simulation experiments.(3)Aiming at the non-Gaussian mixed observation noise distribution with missing statistical information,and studying the target tracking problem under this distribution,this thesis proposes a time-varying particle filter tracking and localization algorithm based on variational Bayesian inference(VB-TVPF).This method first uses Gaussian noise and Rayleigh noise to model non-Gaussian mixed observation noise,then uses variational Bayesian inference to estimate the unknown parameters in the observational noise model,and finally the nonGaussian noise parameters estimated by variational Bayesian are applied to the particle filter to obtain the tracking trajectory of the target.Simulation experiments show that,compared with the classic particle filter algorithm,the proposed algorithm has superior performance and better robustness in target tracking and localization.