Particle Filter Based Track Before Detect Algorithm For Weak Target

With the rapid development of stealth aircrafts in modern war,the detection performance of airborne early-warning radar is greatly affected.The radar system has to face the challenge of detecting and tracking weak target signal in heavy clutter.The particle filter(PF)based track-before-detect(TBD)algorithm use the raw return data or low-thresholded return data to detect and track weak target.The PF-TBD method can estimate the target state in a recursive manner and hence is increasingly concerned in the radar data processing field.Consequently,it has important theoretical significance and military value.In order to solve the problem of detecting and tracking weak target in heavy clutter,several PF-TBD algorithms are proposed in this thesis.The main contributions are given as follows:Firstly,the theoretical principle of PF-TBD is introduced,as well as the dynamic model and measurement model.The detailed implement of PF-TBD algorithm is summarized.In addition,the performance of PF-TBD is evaluated by several performance indices,such as the detection probability and the accept probability of false tracks.Secondly,in order to solve the problem of non-uniform distribution and diversity degradation in PF-TBD,a Quasi Monte Carlo Intelligent PF-TBD(QIPF-TBD)method is proposed.In this method,the initial and proposed particles are sampled by the Quasi Monte Carlo method,which helps to generate a more uniform distribution.In addition,the low-weight particles are re-sampled by the crossover and mutation operation.Simulation results show that the proposed method can effectively improve the particle distribution and particle diversity,and hence improve the detection performance.Finally,a variable rate QIPF-TBD(VRQIPF-TBD)method is proposed to detect and track maneuvering low-SNR targets.In the VRQIPF-TBD method,the target state are updated in a variable rate framework,which sampled more state points in the maneuvering period and less in the smooth regions of target trajectory.The VRQIPF-TBD can decrease the computational complexity and the storage burden in non-maneuvering periods.Simulation results verify the effectiveness of the proposed method.