Research On Heterogeneous Multi-sensor Collaborative Tracking Technique Based On Renyi Information Gain

In military,transportation,industry and many other fields,multi-sensor collaborative tracking technique has been widely used.Multi-sensor collaborative tracking is a technique based on sensor management technique,aiming at realizing the optimal performance of the whole sensor network.It adapts some sensor management model,allocates optimal sensor combination for each target at different moments,and finally accomplishes all targets'tracking task in the monitor area.Compared with homogeneous sensors,heterogeneous sensors can play a more sufficient role in complementary advantages of application scenarios and performance,and achieve a better tracking performance.In this thesis,the main work is as follows:Firstly,on the basis of consulting a large number of related literatures,current research status about heterogeneous multi-sensor collaborative tracking is introduced from three aspects.Secondly,the nonlinear filtering problem is closely related to tracking accuracy and stability of maneuvering target.With regard to its related research,in DMCKF,diagonalization of the covariance matrix is used to substitute Cholesky decomposition in standard CKF to acquire the arithmetic square root matrix and improve the filtering accuracy.However,filtering termination may be caused by the loss of positive definity in the filtering process.Based on calculating the nearest semi-definite matrix of the covirance matrix at every observation moments,an improved DMCKF is proposed to improve the stability and prevent the filtering process from being terminated.At the same time,based on the improved DMCKF,under the Centralized Measurement Fusion(CMF)and the Distributed State Fusion(DSF)architecture,simulation results and the applicable situation of the heterogeneous multi-sensor fusion algorithm are analyzed.Thirdly,considering the key problem of heterogeneous multi-sensor management:heterogeneous multi-sensor multi-target assignment problem,a heterogeneous multi-sensor management algorithm based on Renyi information gain is proposed.The algorithm calculates Renyi information gain through the improved DMCKF.Then the heterogeneous multi-sensor management model is constructed to assign heterogeneous sensor combinations to each maneuvering target at every moment.Fourthly,combining the improved DMCKF with the heterogeneous multi-sensor fusion algorithm based on the improved DMCKF and the heterogeneous multi-sensor management algorithm,a heterogeneous multi-sensor multi-target collaborative tracking algorithm is proposed.The algorithm obtains the fusion of the observed value by adapting the heterogeneous multi-sensor data fusion algorithm based on the improved DMCKF according to the allocation results of sensor resources.Then multi-target collaborative tracking is solved by adapting the improved DMCKF in the IMM framework.After applying the similar process of calculating nearest semi-definite matrix to the standard CKF and UKF,the simulation results show that the improved DMCKF achieves higher collaborative tracking accuracy beyond the former two algorithms.Morever,the improved DMCKF,CKF and UKF perform better than the original algorithms.Finally,the work and shortages are summarized and some problems that needs further study in the future are discussed.