Passive Multi-target Detection And Tracking Based On Multi-static TDOA

In today’s complex electronic warfare environment, the active detection is under threats of the electronic interference and anti-radiation weapons and so on, so passive detection is increasingly important. In this thesis, the radiation source detection technology of passive multi-static TDOA is used as the study object. With the development of electronic technology and the pushing of demand, the issue of detecting and tracking dim targets in low SNR is getting more and more attention. In this case, Traditional detect-before-track(DBT) is faced with problems. Track-before-detect(TBD) directly uses raw measurements, insdead applys a decision threshold to a single measurements,and accumulate target infomation to accomplish detection and tracking, and improve the detection performance.Firstly, the thesis introduces particle filter based on bayesian filter, builds infrared and radar TBD models, then presents two particle filter TBD(PF-TBD) algorithms. The paper has built TDOA TBD model in passive TDOA scene and has chosen the unnormalized weight TBD algorithm as its research direction.Secondly, aiming at the large complexity problem of real-time implementation of TBD algorithm based on particle filter for single target, a TBD algorithm based on BOX particle filter(BPF-TBD) has been proposed. BOX pariticles based on interval analysis replace traditonal spot particles in order to decrease particle number and improve computing speed ensuring detect and track performance. Also, the efficiency of BPF-TBD has been proved by simulation experiment. Using TBD algorithm evaluation criteria, the thesis has made comparisions between the two above algorithms.Finally, as the computational burden of particle filter multi-target multi-bernulli(MeMBer) TBD, a MeMBer TBD based on BOX particle has been proposed. It solves multi-target association and tracks dim targets, and more important use less particles and less time. Simulation experiments are made to compare and verify the effectiveness of the proposed multi-bernulli TBD algorithm based on BOX particle.