Study On Constant False Alarm Rate Detection Approaches For Airborne Radar

The constant false alarm rate (CFAR) detection in the presence of interfering background is always the hotspot and difficulty on the study of radar signal processing. Since the airborne radar adopts the multiple modes and confronts with complicated land and sea clutter, so the CFAR detection for airborne radar is especially difficult. Based on a certain airborne radar system, the theory and technique of CFAR detection for airborne radar in the presence of land and sea clutter environment are investigated in this dissertation. The main works of this dissertation can be summarized as follows:First, the distribution parameter estimation algorithm is studied, since it is the foundation of the optimal or quasi-optimal CFAR detection. In view of the deficiency of extremely time consuming of the maximum likelihood (ML) estimation for K-distribution parameter, the existing seven parameter estimation algorithms based on method of moment (MOM) are analyzed by comparing their estimation accuracy and implementation complexity, and the relations of them are proved, then, a new parameter estimation algorithm based on z~rlog(z) expectation for K-distribution is proposed. The proposed algorithm applies both log transformation and fractional order moment, so its estimation precision is improved evidently. At the same time, a scheme of application of parameter estimation algorithms to airborne radar CFAR detection is presented.Next, the methods of distribution model identification are studied in this dissertation. Aiming at the deficiencies that the common Chi-Square test is affected badly by sample interval division, and the KS test has poor universality, a new clutter distribution identification method, namely, PDF transform method is proposed. The proposed method has the virtues of high identification accuracy and realization simplification of the normal test, moreover, as long as the analytic expression of the distribution model is known, the PDF transform for the clutter sample could be done, so both the identification accuracy and universality are improved evidently. And then, the similarity of four distributions is analyzed by their Skewness and Kurtosis characteristics, and it offers an important academic gist for the precision analysis of the distribution model identification.Thirdly, in the presence of ground clutter environment which the distribution model is uncertain or varying, in view of the deficiency of larger signal-to-noise ratio(SCR) loss with the CFAR detector based on single distribution model, a CFAR detector called CT-CFAR detector which adapt to multi-distribution clutter environment is proposed. The new CFAR detector applies the distribution information of clutter well, so it has better controlling capability of false alarm peak, simultaneity, it achieves certain CFAR gain compared to that for the general rayleigh-based CFAR detector in low SCR. And then, aiming at the deficiencies that the detection performanc of VI-CFAR detector may be seriously degraded when the interfering targets are present in both the halves of the reference window, a new adaptive CFAR detector, i.e. OSVI-CFAR detector is proposed, combining with OS algorithm, this detector achieves better detection performance and false alarm peak controlling capability in multiple interfering targets, clutter edge and homogeneous environment respectively, so it is more robust in practice.Fourthly, the strong correlation of sea clutter has an important influence upon the CFAR detection performance. Based on rational hypothesis of the clutter correlation model, the spatial distribution characteristics of the false alarm probability and detection probability of CA-CFAR is analysed in this dissertation when the spatially correlation length of the clutter structure component is varying. Theoretical analysis and simulation results indicate that the reference window length of CA-CFAR detector should adapt to the change of clutter spatially correlation length when it approaches the optimal detection. But in practical CFAR process, the spatially correlation length of the structure component needs to be estimated. In view of extremely time consuming of the correlation coefficient fitting estimation method in amplitude domain, an approximate method is applied to simplify it. Simulation results prove the efficiency of the approximate method.Finally, the detection performance of CT-CFAR and OSVI-CFAR which proposed in chapter 5 is analyzed basd on measured data. Experiment results demonstrate the correctness and validity of the CFAR detection theory proposed in chapter 5.