Study On Track-before-detect Methods For Radar Dim Target

How to improve the detection performance of radar dim targets and promote their early warning time is one of the great challenges that modern radar are facing. Track-before-detect(TBD) technology is an effective technique for detection and tracking of dim targets. The detection performance of radar dim targets is improved significantly through long time accumulation of multi frame radar target echo, and it has become a hot topic of scholars at home and abroad. Based on discussion of the TBD Methods for Radar Dim Target, this dissertation focuses on the dynamic programming(DP)-based TBD algorithm for high(medium) pulse repetition frequency radars, the TBD method based on range-Doppler extension for radar dim target with range-Doppler ambiguity, the multi-frame detection(MFD) method based on grid partition, the particle filtering(PF)-based TBD method for multi-target detection and tracking in multiple-PRFs radar systems. The main content of this dissertation is summarized as follows.The first part studies a DP-TBD algorithm for high pulse repetition frequency(HPRF) radars with range ambiguity. By considering the ambiguous radial range of the target and range ambiguous number as target states, the problem of range ambiguity resolving is converted into the hybrid state estimation problem. We present a DP-TBD method for joint maximum a posteriori(MAP) estimation of the target trajectory in ambiguous range maps and the corresponding sequence of incremental quantity of range ambiguous number. With the existence of range ambiguity, the detection of dim target through the accumulation of corresponding energy is realized by the proposed algorithm. The simulation result verifies the effectiveness of the proposed method. Finally, a similar method is proposed for medium pulse repetition frequency(MPRF) radars with range-Doppler ambiguity.The second part studies a novel TBD method based on range-Doppler extension for radar dim target with range-Doppler ambiguity. The proposed method functioning in staggering multiple pulse repetition frequencies(PRFs) mode preprocesses the MTD-processed data and range-extends it according to maximum range and Doppler. The maximum energy of targets is then acquired through the accumulation of energy inside the group of multiple PRFs and of track energy between groups. Finally, the track diffusion effect incurred by dynamic programming algorithm is eliminated according to the track cohesion strategy and the track of fake targets by low signal to noise ratio according to the correlation between fake and real targets. The proposed method improves the detection performance of dim target without a priori knowledge of the state of target. The simulation result verifies the effectiveness of the proposed method.The third part focuses on the PF-TBD method for multi-target detection and tracking in multiple-PRFs radar systems. The method uses targets particle swarm technology, which turns the problem of multi-targets detection and tracking into the problem of multiple independent detection and tracking of a single target. Using " Double particle filter" processing structure, the new target is detected and the existing target is tracked. In order to reduce the tracking load of "Tracking layer", the track of fake targets is eliminated in "Discovery layer". The particle swarm and track of targets is maintained or deleted in "Tracking layer". A new ESIR-TBD particle filter method is proposed to resolve the ambiguity problem of range-Doppler cells, and a probabilistic data association method is proposed to realize track association of close distributed target. In the proposed algorithm, the number of target is estimated by the particle swarms, and the state of targets is estimated by corresponding probabilities of target existence. The simulation results show that in the multiple-PRFs operation mode, the detection and tracking of multiple targets are realized effectively by the proposed method.The fourth part studies the MFD method based on grid partition. It does not require a discretization of the state space and operates directly on the input plot-lists. The proposed architecture consists of a pre-processing stage, which extracts a set of candidate alarms from the raw data measurements. And then the set of candidate alarms is divided into some grid partition, by setting the appropriate size of grid partition and using the Doppler information of the target, the searching range of target is limited in some grid partition around the target. Finally, a TBD processor is used to jointly elaborate observations from multiple frames candidate alarms and confirms reliable target. A computationally efficient dynamic programming algorithm for the TBD processor is derived, which is the combination of sliding window track energy accumulation and successive track cancellation(STC), target detection delay is overcome, and interference of closely-space targets is to be eliminated at the same time. Simulation results show that, compared with the existing methods, this method improves the detection performance of the radar dim targets, reducing the computational complexity of the algorithm.