Research On Adaptive Subspace Signal Detection Method

Adaptive detection is an important research subject in the field of signal processing,especially in the field of radar signal processing.Traditional adaptive signal detection methods are based on the rank-one signal model to match the target echo signal.However,these detection methods do not consider the uncertainty of the steering vector,and it is easy to cause serious degradation of detection performance due to the mismatch of the steering vector.In addition,the use of the training sample for clutter covariance matrix estimation is a necessary step of adaptive signal detection,but in the actual target detection scenario,available independent identical distribution training sample data is limited,at this time,the clutter covariance matrix cannot be estimated accurately,thereby seriously affecting the performance of the detection method.Given the above problems,this thesis adopts a multi-rank subspace model to constrain the actual steering vector of the signal,and combines the persymmetric structure information of the clutter covariance matrix,then designs the adaptive subspace signal detection methods based on GLRT,Rao,and Wald criteria to enhance the detection performance of the methods and their robustness against the uncertainty of the steering vector.The main work of this thesis is as follows:1.The echo signal of radar clutter is modeled,and an adaptive subspace signal detection method for point-like target based on covariance structure information under compoundGaussian clutter background is proposed,and finally the effectiveness of the proposed detection methods is verified by simulation experiments,and especially when the training data is insufficient,compared with the traditional detection methods,the superiority of the proposed detection method is demonstrated.2.Based on the multi-rank subspace model and the persymmetric structure information of clutter covariance matrix in space or time central symmetric detection scenario of radar receiving unit,a multi-rank range-spread target detection method for non-Gaussian clutter background of multi-channel array radar is proposed.The constant false alarm rate property of the proposed detection methods with regard to the covariance matrix of clutter is illustrated by theoretical analysis,and the effective detection performance of the proposed methods in the case of insufficient training samples and the robustness of the proposed method to steering vector mismatch is verified by simulation data and measured clutter data respectively.3.Based on Rao and Wald test criterion,respectively,puts forward a Uo S detection method,and derive the detecting statistics and classification probability boundary.Simulation experiments validate the proposed methods have better detection performance.4.A method of Uo S signal detection based on covariance structure information are proposed under a partially homogeneous clutter background.Simulation experiments verify the constant false alarm rate property of the proposed methods with regards to the clutter covariance matrix and indicate that compared with traditional detection methods,the proposed methods present higher detection probability in small sample scenarios,and can determine the active subspace of the target more accurately.