Research On Track-before-detect Algorithms For Weak Targets

With the development and wide utilization of stealthy technology, the radar crosssection (RCS) of typical military targets, such as aircrafts, missels and combat ships,has been greatly decreased. The reflected signal from these targets is very weak causingsevere challenges to radar detection and tracking. The traditional Detect-Before-Track(DBT) methods cannot guarantee reliable detection and tracking performance whenencounted with these weak targets. Track-Before-Detect (TBD) methods are one of theeffective methods of detection and tracking weak targets, which jointly process severalconsecutive scans of unthresholded data to integrate the targets energy, jointly declearthe presence of the targets, and, enentually, its track. As a developing new technique,it’s urgent to investigate TBD theories, modifiy TBD methodologies and expend TBDapplication for weak targets in radar system which can improve detection and trackingperformances.In this dissertation, TBD algorithm for weak targets in radar system is investigated.The main results are as follows:1. The differences and the advantages and disadvantages of DBT methods andTBD methods are analysised and comparisoned. And TBD methods are classified ascoherent accumulation and non-coherent accumulation among the frames by the way ofenergy accumulation. The signal processing process and algorithms of the two ways arepresented, respectively. This chapter makes the theoretical foundation of the subsequentchapters of this thesis.2. A modified polar random Hough transform based TBD algorithm (MP-RHT-TBD) is proposed. The criterion of maximum sampling number is presented. Thetargets’ motion information among frames is utilized to restrain the amount of invalidsampling, and the rule of voting by minimum distance overcomes the problem of peakbroadening. The MP-RHT-TBD algorithm has better detection and trackingperformaces than the P-RHT-TBD algorithm.3. A particle filter based TBD algorithm for range extent targets is proposed. Themotion model and measurement model of multi-scatterers for range extent targets are built. A particle filter based TBD algorithm for range extent targets is proposed toestimate the presence/absence, the motion states and length of the range extent targets,simultaneously. By comparison with the existing TBD algorithm for extent targets, theproposed algorithm has better performance of the detection and tracking.4. A novel PF-TBD based on adaptive Markov Chain Monte Carlo (MCMC) formulti-targets is proposed. By the adaptive sampling strategy, adjacent targets are jointsampled and far-away targets are independent sampled. This adaptive samplingimproves efficiency of algorithm and has better rate of convergence than MCMC-PF-TBD algorithm, and has better tracking precision than the SIR-PF-TBD algorithm.5. A multi-frame coherent TBD technology is proposed. By the analysis of theechoes of the moving targets, a signal model for coherent accumulation among frames isbuilt. Based on the model, the GLRT detector is derived, and an efficient coherentaccumulation algorithm is proposed to estimate the unknown parameters of the target(targets’ azimuth, distance, Doppler and modulation slope). The analytical expressionsof false alarm and detection probabilitiy of GLRT detector based on the proposedalgorithm are derived, and the performance of detection of the algorithm is analyzedtheoretically. Finally, the validity of the signal model and the efficiency of the proposedalgorithm are verified by simulation experiments.