Studies On Target Localization And Tracking In Passive Coherent Location Radar

As a new type of radar, external illuminator based passive radar (also namedpassive coherent location radar) is a variation of bistatic radar that exploitsnon-cooperative illuminators of opportunity (such as FM broadcast, television, GSM,and GPS transmitters, etc) as its radar transmitters, and it is to detect target by onlyreceiving the echo signal reflected by the target. Unlike the traditional radar, its receiveris absolute passive, making it far less vulnerable to electronic counter measures andanti-radiation guided missile, so this radar may have better survival in a hostileenvironment. In addition, PCL radar has better ability to detect low altitude target andstealth target by right of the signal characteristic of FM, television, etc. Therefore, PCLradar has become one of the most effective approaches to detect target undercomplicated electromagnetism environment, and it is valuable to study.This dissertation addresses some aspects of PCL Radar for target localization andtracking. The research focus on the target localization precision, the influence of theconfiguration of PCL radar on target localization and tracking, the influence of glintnoise on target tracking, nonlinear and non-Gaussian filtering problem, andmaneuvering target tracking.The main content of this dissertation is summarized as follows.In the first part, the concept, the background, and the study value of PCL radar areintroduced, an overview of the development history and the available tracking methodsfor PCL radar is given. Then, we brief introduce the signal processing flow, the keytechnologies of the PCL radar, and the basic location methods. We also derive theKalman filter and the Extended Kalman filter based on the Bayesian theorem, and pointout the puzzle in PCL radar localization and tracking. At last, the particle filter-basedidea for target tracking of PCL radar is proposed, which form the basis of the followingstudy.In the second part, the accuracy of target localization in PCL radar is studied. First,the precision of measurements parameter is analyzed, and the factors which affect theprecision of measurements parameter are pointed out. Then, geometrical dilution ofprecision (GDOP) of the target relative coordinate errors are formulated. The influenceof the target location method, the mode of location, and the configuration of PCL radaron the GDOP is analyzed. Finally, the location method and the mode of location whichsuit for PCL radar are found out.In the third part, the tracking algorithm for PCL radar in the presence of non-Gaussian noise (glint noise) is studied. In PCL radar, the tracking performance ofEKF is affected seriously by the glint noise. To solve this problem, a new passive radartracking method is proposed based on particle filter (PF) and time of arrival (TOA)measurements. The method gains TOA measurements by multi-stations, utilizes aGaussian distribution and a student distribution to construct the glint distribution andthen uses nonlinear and non-Gaussian PF to track target, which avoids the error causedby EKF linearization. The simulations show that the new method overcomes the EKF,especially in glint noise environment. The real data farther demonstrates the validity ofthe proposed method.In the fourth part, an improved particle filtering based on differential evolution(DE) algorithm is proposed. The traditional resampling scheme results in theimpoverishment phenomenon. To deal with this problem, the evolution idea isintroduced to the resampling scheme in particle filtering, and a new resampling schemebased on the DE algorithm is presented, in which the resampling process is regarded asa process for searching better particles. By using this scheme, an differential evolutionparticle filter is proposed, and then it is applied to the target tracking in PCL radar. Theperformance of the new PF and the standard PF is analyzed by the simulations, and thefactors affect the new PF are also analyzed. Simulation results demonstrate that theproposed PF outperforms the standard PFIn the fifth part, the differential evolution particle filter is further studied. Anunscented Kalman filter (UKF) is used to generate the importance proposal distribution(IPD) of particle filter, which matches the true posterior more closely and canincorporate the latest observations into a prior updating routine. In addition, fourdifferential aberrance operators are used in the DE resampling scheme, and four DEresampling schemes are presented. As a result, four types of differential evolutionparticle filters (DEPFs) are proposed, in which the UKF is utilised to generate the IPDand the DE resampling schemes are used as the resampling scheme. Simulation resultsdemonstrate that the proposed DEPFs outperform the basic differential evolutionparticle filter, and the unscented particle filter.In the sixth part, the problem of manoeuvring target tracking in PCL radar withglint noise is studied. The dynamic state space (DSS) model of manoeuvring target isconstructed. To deal with the problem of manoeuvring target tracking in the presence ofglint noise, an interacting multiple model (IMM) particle filtering method usingmultiple TOA measurements from several transmitter-receiver pairs is proposed andevaluated. The influence of glint noise and configuration of transmitters and receiver stations on the tracking method are analyzed. Simulations illustrate that, compared tothe IMM and the standard PF, the proposed method obtains better estimates of position,velocity, and acceleration, and is smaller affected by the glint noise and configuration ofthe PCL radar.