Radar CFAR Detection And Distributed Radar CFAR Detection Under Complicated Environments

After its development through the past years,radar constant false alarm rate (CFAR) detection becomes a standard technique in modern radar design. Now a series of principles,conceptions and approaches have been established for CFAR detection,and they are widely used in varies type of radars. But been a technical of signal detection,CFAR are far from consummate,not only in practice but also theoretically,particularly for the target detection in interference and non-Gaussian environments,which are typically in practical situations. Distributed CFAR detection have has been gaining importance accomplished with the widely concerns on distributed radar systems. The basic theories of distributed CFAR detect systems,such as the system topology,the hypothesis testing with data fusion,the performance estimation and fusion rule,are established by the works of many researchers,and most of the concepts and approaches of traditional CFAR detection can be used on distributed systems. But when worked on practical environments with unstable and complicated detect conditions,the distributed system have many distinctly properties compared with single detector,so there exist a great deal of problems need to be studied.This dissertation is based on the works of former researchers,and we are concerned with the principles,approaches,performance properties and the design rules of radar CFAR detection and distributed radar CFAR detection systems under complicated environments (such as interference,clutter and varied environments with non-identically detect conditions). The main contributions of this paper are summarized as follows:(1) The basic results on radar CFAR detection are reviewed. In particular,the CFAR detection under nonhomogeneous backgrounds are summarized and analyzed in some details. Using the numerical performance simulation results,the performance of mean level (ML),order statistic (OS) and trimmed mean (TM) type CFAR detectors are analyzed and compared in varies conditions. For the proportion of practical application and further development of radar CFAR techniques,we have summarized the performance characteristics of these detectors.(2) Based on the work of former mentioned,we have proposed a new CFAR detect method which is called cell select (CS). The detection and false alarm performance of this detector is derived and compared to several previously proposed CFARs. The results indicate that the CS can improve the performance of CA and OS-CFAR detectors under interference environments obviously,without any performance losses when the background is homogeneously. But in an environment with edge clutter,CS provides worse protection against loss of false alarm control. So the greatest-of (GO) selection of dual reference windows is introduced to CS detectors for the purpose of improve its false alarm performance on edge clutter environments. The resulted CFAR detector which posses CS and GO selection processor is proved to be one of the best CFAR detector we have encountered,it can provides perfect performance on interference environment and good false alarm control property on edge clutter backgrounds.( 3) The basic theories and techniques on distributed radar CFAR detection are reviewed. In particular,using the numerical performance simulation results,the performance of distributed CFAR detect system under nonhomogeneous backgrounds and simple detect conditions are analyzed in some details. We have summarized the performance characteristics of distributed CFAR detect systems against interference and edge clutter when different fusion rule is used.(4) Based on the work of the third point,we have first worked on the system optimization and performance analyses of distributed OS-CFAR detection system under complicated detect conditions (which means a system with more than two sensors and each sensor works on different conditions). Using simple genetic algorithms and two modified genetic algorithms named simulated annealing genetic algorithms and niche genetic algorithms,a typical distributed OS-CFAR dete...