Study Of Bispectral Signatures Of Conventional Radar Targets And Its Applications

ABSTRACTThis paper mainly deals with the bispectrum of complex-valued signals analysis techniques as well as their applications to conventional radar target signature analysis. Firstly, the concepts and properties of high-order statistics are introduced with emphasis on the features of the high-order statistics of a Gaussian process and the illustration of the advantages of bispectrum analysis applying to radar signal analysis. Secondly, according to the characteristics of conventional coherent radar echoes, the definition and properties of the bispectrum of complex-valued signals are studied together with the development of the parametric bispectrum estimation methods. On the basis of this, by analyzing the actual conventional radar target return data with the methods, the target bispectral signature are investigated and a method for target bispectral signature selection and template construction is proposed. By using a minimum distance classifier, a target classification is performed with some satisfactory results. Finally, the influence of white and color Gaussian noise and ground clutter on the performance of target classification are discussed, and some experiments show that the bispectrum analysis is able to reject the noise with Gaussian distribution and clutter. A conclusion can be made that the application of bispectrum analysis techniques in radar target return analysis is effective and practical.