Airborne Radar Space-time Adaptive Processing Based On Clutter Structure

The clutter environment faced by airborne radar is very complex.How to suppress clutter has always been the main problem of the airborne radar.Space-time adaptive processing(STAP)utilizes the coupling characteristics of clutter,suppresses clutter by forming notches at the clutter locations,and is optimal under conditions where the clutter covariance matrix of the cell under test(CUT)is known.However,in the actual environment,the clutter covariance matrix of the CUT is usually unknown.It needs to be estimated by tranning samples.These tranning samples must satisfy the conditions of independent and identical distribution(IID).In order to make the SNR loss within 3d B of the optimal STAP,the number of training samples is required to be twice the degree of freedom(DOF).When the DOF is high,the number of the training sample needs to be very large.It is difficult to be realized in practical applications.First because of it is difficult to obtain enough IID training samples in nonhomogeneous enviroments.Second is the computational complexity will be vary high.This thesis focuses on this issue.First,this thesis introduces a series of partially space-time STAP algorithms,such as m DT,JDL,PC,etc.This kind of algorithm uses dimension reduction or rank reduction to convert fully adaptive processing into partially one,which could not only reduces the complexity of computation but also reduces the number of training samples needed in fully STAP.This thesis combines the two methods to further improve the performance of such algorithms when the number of training samples is small.Second,there are two problems with the above-mentioned kind of statistical algorithm.Firstly,the performance can be significantly reduced when there are non-homogeneous components in the data.Secondly,this kind of algorithm can't solve the problem that the CUT has non-homogeneity.For example,when there is an interference in the CUT,because the training sample selected by the above algorithm does not contain the information of the interference,it cannot be effectively suppressed.For the first point,this thesis introduces two methods,the non-homogeneous detection algorithm generalized inner product and the spectrum-based sample selection method.For the second point,this thesis introduces a kind of method based on the generalized sidelobe cancellation structure.Finally,When the radar system is equipped with some certain structures,the covariance matrix of the received clutter will have some special structural characteristics.For example,the clutter covariance matrix has a Hermitian persymmetric(also called centro-hermitian)form,when the used system is equipped with a symmetrically spaced linear array or symmetrically spaced pulse trains.Hermitian persymmetry has a property of doubly symmetry,i.e.,Hermitian about its principal diagonal and persymmetric about its cross diagonal.In this thesis,the centro-hermitian is added to the traditional principal component method and cross-spectral scale method to improve the performance when the number of training samples is insufficient.In addition,this thesis combines the structural characteristics of the clutter,the principal component method and the traditional adaptive matched filtering method,proposed a reduced rank adaptive matched filtering method based on the centro-hermitian.