Distributed Track-to-track Fusion Method With Passive Multistatic Radar

Passive multistatic radar(PMR),also known as passive coherent location(PCL),uses frequency modulation,digital television,GPS,GSM and other cooperative or non-cooperative transmitters signals to detect the target.Since the PMR does not need transmitting antenna itself,it has the following merits:silent,low cost and flexible installation.More importantly,the PMR can takes advantage of the spatial diversity and frequency diversity to detect stealth targets,and has received increasing attentions in the radar field.When multiple receivers are used in a PMR system,the quality of target track can be further improved by information fusion technology.In this paper,we focus on the problem of distributed track-to-track fusion in the presence of clutter using the PMR with multiple receivers.The main research is given as follows:Firstly,in order to fuse multiple tracks in clutter-free environment,several classical track-to-track fusion methods,such as covariance convex combination,information matrix fusion,covariance interaction,is analyzed systematically.Simulation results show that the covariance convex combination method is simple and feasible,but it ignores the correlation of local estimation.The information matrix fusion method,however,uses the prediction information of local receiver to improve the tracking accuracy of fused track in part.The covariance interaction fusion method can effectively improve the track quality by weighted fusion of local estimation error cross covariance.Secondly,in order to solve the problem of PMR track-to-track fusion where the cross covariance of local estimation error is unknown,a fast covariance interaction fusion method with memory and a fast fault-tolerant generalized convex combination fusion method with memory are proposed respectively.Through the feedback of the fusion state from the previous frame,the former obtains the closed solution of the weight value based on the information theory optimization criterion.The latter,under the same framework of the former,introduces an adaptive weight parameter to optimize the weight,and further improves the tracking accuracy of the fused track.Simulation results verify the effectiveness of the proposed two methods.Finally,in order to solve the problem of track-to-track fusion in clutter where local estimation error cross covariance is unknown,two integrated generalized convex combined track-to-track fusion methods considering the existence probability of the target are proposed to improve the probability of correct track and tracking accuracy.The first method is named as the integrated generalized convex combination fusion method without memory.It obtains the candidate track set and the corresponding target existence probability with a single receiver based on the integrated probabilistic data association method,and then embeds the target existence probability into the track-to-track fusion process to improve the fusion track quality.The second method is named as the integrated generalized convex combination fusion method with memory.It is obtained by using the feedback of the fusion state from the previous frame.The simulation verifies the validity of the proposed method.