Research On Radar Target Detection And Tracking In Compressive Measurement Domain

In order to detect and track the targets more effectively,modern radar system usually needs to adopt a large bandwidth and multi-channel signal processing method to improve the detection performance of target parameters(distance,speed,angle,etc.)such as the maximum detection range,resolution and measuring accuracy.Limited by the Nyquist sampling theorem,large bandwidth is often accompanied by high sampling rate and large amount of data and so on.However,targets are highly sparse in the radar detection background,and the effective targets information only accounts for a very small part of the massive radar data,which results in the waste of resources and low efficiency of the radar system in realizing target detection.In recent years,the Compressive Sensing(CS)theory provides the possibility to solve the above problems from the perspective of signal sampling.Compressive Sensing Radar(CSR)can greatly reduce the quantity of sampled data and guarantee the integrity of the targets information simultaneously.However,the traditional CSR just transfer the stress from signal acquisition to signal processing,that is,it needs complicated signal reconstruction algorithm to achieve the target detection and tracking,which is contrary to reducing the complexity of the radar system and improving the real-time performance of the system.Thus,scholars studied the the second category of CSR by directly dealing the measurement.The CSR in Compressive Samping and Processing(CSP)framework,without signal reconstructing,complete target detection,recognition,tracking and classification task directly in the compressed measurement domain.It makes full use of the CS theory in radar signal acquisition,transmission and processing.However,in the CSP framework,the low signal-to-noise ratio(SNR)caused by compressive measurement seriously restricts the performance of CSR.Low SNR may cause a series of CSR problems,such as the difficulty of compressive sampling,the performance degradation of the compressive sampling operator,and the inability of the compressive detector to detect the target.As the core of CS theory and an important designable module of CSR,the design of compressive sampling operator is an important way to solve the above problems.To improve the radar target detection and tracking performance in compressive measurement domain,this dissertation combines "radar signal processing with compressive measurement" and "optimization of target detection and tracking".It takes the effective detection of radar target in compressive measurement domain as the research goal,and adoptes signal subspace estimation with target tracking results and the optimal compressive sampling operator design as the main technical methods.The main contributions and innovations of this dissertation are as follows:(1)A Pulse Repetition Rate Compressive Sampler(PRRCS)is studied.To solve the problem that the existing high rate mixing circuit of Analog to Information Converter(AIC)is not easy to realize in engineering,a PRRCS is proposed by using the strong correlation of sequent radar pulses.This AIC makes full use of the sparse characteristics of radar signals in fast time range dimension and slow time pulse dimension,and designs a compressive sampling operator with better structure.Compared with Random Demodulation(RD)which only samples a single pulse,the proposed AIC is easier to implement and has a higher data compression rate.(2)The step-by-step optimization method of single target compressive detection and tracking is studied.Firstly,a pre-tracking compressive detector for single target is studied to solve the performance degradation problem caused by low SNR.Based on the target tracking results,the optimal compressive sampling operator is designed to optimize the compressive sampling and compressive detection at the next moment.Secondly,the proposed mean-shift positioning method is used to locate the specific target position.Finally,a multi model filtering algorithm is proposed by establishing a variety of measurement noise models to solve the problem of slow time-varying compressive measurement noise,and an adaptive Kalman filtering algorithm is proposed to estimate the fast time-varying compressive measurement noise approximately by using the Variational Bayesian method.This study followed the idea of tracking before detection,and realized the step by step optimization of single target detection and tracking,the experimental results verify the excellent performace of these methods.(3)The joint optimization method of single target compressive detection and tracking is studied.To solve the problem that the global optimal solution of the system cannot be found by single or step-by-step optimization of detection and tracking,a joint optimization method of single target compressive detection and tracking is proposed.This method introduces the optimal bayesian Joint Decision and Estimation(JDE)theory into the CSP framework,deduces a new joint optimization risk function,and gives the optimal solution of the function.The method adopts double closed-loop feedback structure,and the target tracking results are used not only to optimize the compressive sampling operator and the compressive detector,but also to optimize the target tracking at the next moment.Experimental results verify the global superiority of the method in joint detection and tracking.(4)The step-by-step optimization method of multi-target compressive detection and tracking is studied.To solve the problems of the large amount of radar data in the existing radar multi-target tracking system and the poor real-time performance of the tracking algorithm based on data association,a step-by-step optimization method of multi-target compressive detection and tracking is proposed.Firstly,the pre-tracking compressive detector and positioning method for single-target were extended to multi-target application scenarios,and then the target parameters obtained by the compressive detector were processed by a Probability Hypothesis Density(PHD)filter.The simulation experiments have demonstrated that this proposed method can improve the real-time and tracking performance of the multi-target tracking system in compressive measurement domain.(5)The joint optimization method of multi-target compressive detection and tracking is studied.Following the idea of joint optimization for single target,the problem of joint optimization for multi-target with known and fixed target number and uncrossed target track is studied.By decomposing the simple multi-target problem into parallel multiple single target problems,a joint optimization for multi-target risk function is derived and the optimal solution of this function is given.This method is an extension of joint optimization for single target.The experimental results verify the global superiority of this method.To sum up,from the perspective of directly dealing with compressive measurement,this dissertation adopts the idea of step-by-step optimization and joint optimization to improve the detection and tracking performance of radar targets in compressive measurement domain.