Research On The Detection And Discrimination Of Unresolved Targets Of The Low Resolution Radar

This dissertation researches on the methods of detection and discrimination of unresolved targets of the low resolution radar.The background is introduced and the development of the technology of detection and discrimination of unresolved targets are briefly reviewed in chapter 1.According to the targets' angles , the methods of detection and discrimination are studied in chapter 2. The probability density function (pdf) and statistics of monopulse ratio for unresolved Rician targets are strictly derived. The formulas for two Rayleigh targets' detection in literature are extended to N targets. The three improved GLRT (generalized likelihood ratio test) methods are proposed by using the two angle measurements of the outputs. Simulation results show that the three methods are better than GLRT. The estimation methods of unresolved targets using some prior information are put forth with unknown targets' SNR (signal-to-noise ratio).Methods are studied by making use of the targets' angles and ranges in chapter 3. Based the data model using two sampling points, the algorithm of detection and estimation using the Gibbs sampler and MDL (minimum description length) criteria is proposed and simulation results show that the algorithm is better .The data model by using three sampling points is set up and based the model 10 targets can be theoretically discriminated.Methods are studied by making use of the target's angle and Doppler frequency in chapter 4. To overcome the shortage of inaccuracy of the disturbance power estimation in literature, an improved adaptive SHT (successive hypotheses testing) is proposed. A detection method different from SHT is given based the fact the new estimation method can estimate the number of targets, which can adaptively set the maximum number of targets. The simulation results prove that the new estimation method has higher accuracy and practicability and the improved SHT is better in detection performance.In chapter 5, the multiple non-Rayleigh targets is researched. The receiver outputs' pdfs of multiple Swerling IV targets are strictly derived. By a lot of simulation, an important conclusion is drawn that the condition under which the outputs' pdfs can be approximated by Gaussian pdfs is that the difference of two targets' SNR which have biggest return power among the targets is not more than 7dB. Under the condition, results show that the detection and estimation of multiple Swerling IV targets is good by using the proposed methods in this dissertation.At last, summary of this dissertation is made and the present problems and further research are pointed out.