Font Size: a A A

Study On Signal Processing And Target Tracking Method For Passive Radar

Posted on:2017-03-24Degree:DoctorType:Dissertation
Country:ChinaCandidate:Y WuFull Text:PDF
GTID:1108330488957287Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
Passive radar (PR) system can be described as a "receiver-only" radar, without a purposely-built transmitter. It exploits the already-existing non-cooperate electromagnetic radiations, known as "illuminator of opportunity", to detect the targets in a specific area. Relying on the characteristics of silent receiving, bistatic configuration and signal band used, PBR has incomparable advantages over conventional radars, and is a very promising radar system. For example, because of no electromagnetic pollution, the PBR can be deployed in urban, suburban or densely populated regions to detect targets.Traditional PR usually utilizes the "one transmitting-one receiving" mode to detect, locate and track the targets. However, as the technology continues to evolve and upgrading, traditional detection model gradually fails to keep pace with the needs of society. Currently, in order to effectively deal with the increasing demands for civilian or military, the PR technology gradually moves towards the road of the development of wideband and multi-source. Compared with the traditional PR, one of the most direct changes in wideband and multi-source mode is that the signal processing scale increases by dozens or even hundreds of times, bringing a no small pressure for the conventional hardware processing ability. On the other hand, the non-cooperation and fluctuation of the illumination source used by PR will cause the detected target track irregular intermittent problem. Refined tracking filtering method is an effective way to make up for the institutional shortcoming. Under the support of some national project, this dissertation focuses on PR signal processing and target tracking filtering methods to carry out the relevant work.The main contents of this dissertation are summarized as follows.1. Several practical issues on conventional PBR signal processing are studied. For practical applications, the shortcomings existed in the conventional signal processing methods are analyzed, and then the corresponding optimized or improved methods are put forward. For instance, a comprehensive target detection method including amplitude correction, unit average, compensation, far-near partition, two-sorting, etc, is proposed to overcome the problem that the Range-Doppler detection platform shows a ramp-shape caused by the near strong remaining clutter. As an another example, for the case where the reference channel may also receive the target echo signal, the mechanisms of the energy of real targets being counteracted and producing false targets are first theoretically analyzed, and then we propose to exploit the coefficients of the weights of clutter suppression to eliminate the target echo signal existed in the reference signal. The experimental results shows that the proposed method can effectively solve the problems mentioned above caused by the target echo signal in the reference signal.2. Real-time processing for multi-source omnidirectional PR signals is studied. Specifically, a general-purpose processor GPU-based parallel implementation method is presented to handle the difficulty of the real-time processing for multi-source omnidirectional PBR signals. Through the multi-level longitudinal optimizations, the processing efficiency of the proposed method is significantly improved, reaching the level of real-time processing in the application. Compared with traditional DSP processing framework, the proposed method has the advantages of low-cost, easy to debug, small-scale hardware and easy to be transferred or extended. In addition, the proposed method is in line with the designing principles of "software radar". Finally, the feasibility and efficiency of the proposed method is verified by the simulations and real data.3. A weighted least-squares particle filter which can be applied to, but not limited to the PBR target tracking is introduced. Specifically, for the unavoidable particle degeneracy problem in the nonlinear particle filter, we propose to employ the least-squares estimate to incorporate the current measurement information into the particle sampling stage so as to guide the particles that lie in the prior regions to move towards the significant areas where true state is most likely to exist, resulting in the improved quality of the sampling. In addition, the measurement term is extended using the state vector to deal with the potential rank-defect problem, thereby guaranteeing the generality of the proposed method. At last, the simulations and real data indicate the validity of the proposed method.4. A parallel resampling (PR) is proposed to cope with the problem that the systematic resampling (SR) restricts the degree of parallelization of particle filter. The basic idea of the proposed PR is similar to that of SR but performed differently. In short, the PR directly used the integral part of the product of the particle weight and particle number as the number of times that a particle was replicated, and it simultaneously eliminated several particles with the smallest weights so as to maintain a constant particle number, which are the two key differences from the SR. The proposed method has the advantages of simple structure and easy to be implemented in parallel.
Keywords/Search Tags:Passive Radar, Target Detection, Real-time Processing, Particle Filter, Resampling
PDF Full Text Request
Related items