Research On Application Of Blind Source Separation In Radar Reconnaissance

In the past 30 years,Blind Source Separation has attracted many scholars' attention,because of its unique advantages.In the case of source signal's number and mixed characteristic are unknown,it only relies on the observation data of array output to restore or extract the source signal.At present,blind source separation is a research hotspot,has a wide range of applications in the field of signal processing,such as in the voice signal processing,medical signals,medical images,communication signal processing,radar signal processing and other aspects.Based on the study of blind source separation algorithm,this paper applies blind source separation to radar signal separation and DOA estimation.The main research work is described as follows:1.Introducing the theory of blind source separation,including blind source separation model,the basic theory of blind source separation,objective function and the optimization algorithm and performance index.And the classical blind source separation algorithm is simulated.At the same time,the basic theory of arrival angle estimation in array signal processing is introduced,and the possibility of combining blind source separation with array signal processing is discussed.2.In the second part,based on the study of classical algorithms,the problem of noisy source separation is discussed.Denoising Source Separation is introduced and applied to radar signal separation based on noisy array model.In this paper,an improved denoising algorithm based on singular value threshold is proposed.The singular value threshold algorithm based on Stein unbiased risk estimation is used to replace the singular value decomposition in the denoising source separation.The simulation results show that the algorithm improves the effect of noisy source separation.3.In the third part,we discuss the application of blind source separation algorithm based on high-order statistics in the DOA estimation in array signal processing.We study an algorithm of DOA estimation based on joint approximation diagonalization and multiple signal classification algorithm,using the high-order statistics of the observed data,then the corresponding vector of the observation data is separated,and then the peak search is carried out for each column of the steering vector to complete the DOA estimation.4.Finally,it points out the possible future research directions in the radar reconnaissance's blind source separation.