Non-coherent CFAR Detection In Sea Clutter For Coherent Radar Systems

Radar is an important mean of monitoring and detecting targets in sea clutter and it has a significant role in civil and military fields.The methods of detecting targets in sea clutter for different radar systems have been an important content in the field of radar.This paper focuses on the non-coherent CFAR detectors in sea clutter for coherent radar systems.In practical application,the non-coherent CFAR detectors are often used in the coherent radar systems.However,the non-coherent detectors will not be CFAR detectors because of the correlation of sea clutter in the whole coherent process interval.Therefore,this paper presents pre-whitening cell averaging detection and pre-whitening cell median detection which are based on the pre-whitening method in the coherent radar systems.Firstly,this paper studies the statistical characteristics and the K distribution model which is a kind of the compound-Gaussian model.The K distribution model is often used as sea clutter model beacause it not only can descibe the characteristics of amplitude well,but also can describe the spatial correlation and temporal correlation of sea clutter.Then,the formation condition of spatial correlation and temporal correlation is analyzed in detail.Next,the structure of non-coherent constant false alarm detector,the mean level CFAR detectors and the ordered statistics CFAR detectors are introduced.Secondly,the covariance matrix estimation methods and the shape parameter estimation methods of K distribution are introduced.The covariance matrix estimation methods include the normalized sample covariance matrix estimation method,the maximum likelihood estimation method and the approximate maximum likelihood estimation method.The maximum likelihood estimation method and the approximate maximum likelihood estimation method have similar performance,and they are complex.The normalized sample covariance matrix estimation method has a simple structure and it can get results in a short time.The estimation methods of shape parameter are the moment estimation method and the bi-percentile method.And the advantages and disadvantages of the two methods are introduced.Finally,this paper introduces the cell averaging detector and cell median detector in coherent radar systems.In this paper,the probability density functions of the cell averaging detector and cell median detector are deduced.And the result shows that the probability density functions are unrelated to the scale parameter and related to the number of reference cells.Then,some simulations are done and the results show that the probability density functions are related to the accumulate pulse number,shape parameter and the covariance matrix.Therefore,this paper presents pre-whitening cell averaging detection and pre-whitening cell median detection which are based on the pre-whitening method.And the thresholds of these two detectors are adaptive.Then,the performance and the calculation time of the two methods proposed are compared.