Airborne Radar Space-time Processing Based On Structural Characteristics Of Covariance Matrix

As modern airborne radar,especially early warning radar is widely used by largescale arrays and multi-pulses.In addition,because of the variable terrain characteristics and the heterogeneity of the airborne radar environment,the large number of samples needed in radar signal processing is in contradiction with the number of samples provided by the environment.In order to solve the problems of the clutter covariance matrix estimation,dimension reduction algorithm,waveform design and target detection in space-time adaptive processing(STAP)of airborne radar with or without reference samples,some prior information is adopted to improve the STAP performance of airborne radar.Therefore,this paper focuses on the structure characteristics of the airborne radar covariance matrix.This paper also focuses on the related space-time two-dimensional signal processing algorithm research.The main works and contributions of this paper are as follows:(1)Establishment of airborne radar target clutter model and analysis of clutter covariance matrix structure.Airborne radar signal modeling is the basis of the follow-up work.Firstly,the target and clutter models of the airborne phased array radar and the multiple-input multipleoutput(MIMO)radar are introduced,respectively.And the corresponding clutter covariance matrices are given.Based on this,it is concluded that the clutter covariance matrix has a certain special structure,including that most covariance matrix structures belong to a set of linear structures.Using this linear structure,the maximum likelihood estimation method of covariance matrix and several scaling estimation algorithms are studied.(2)Estimation of the structured clutter covariance matrix for airborne phased array radar.The clutter covariance matrix of airborne phased array radar has the non-linear structure with summing up Kronecker products of multiple low-rank spatial and temporal clutter basis matrices.Resorting to a Kronecker product permutation operation,the nonlinear structure can be transformed into a linear structure by the sum of vector outer products.Based on this,with the help of proximal gradient algorithm,this paper proposes an iterative estimation algorithm which combines maximum likelihood estimation,the prior covariance matrix and the rank constraint.It is theoretically proved that both the covariance matrix and the basis matrices obtained by the proposed estimation algorithm are Hermitian matrices.(3)Two-stage STAP and covariance matrix estimation of airborne MIMO radar based on the structured clutter covariance matrix.The clutter covariance matrix of airborne MIMO radar can be expressed as a nonlinear structure with summation of Kronecker products by basis matrices in transmitting,receiving spatial and temporal domains.Based on the structured weight vector with the Kronecker product structure,a processing mode of the two-stage STAP is proposed in this paper,which can effectively reduce the computational complexity under the condition of superior performance of the output signal-to-clutter-noise ratio(SCNR).In addition,to solve the clutter covariance matrix estimation problem,a structured clutter covariance matrix estimation algorithm based on the two uclear norms is proposed,which can effectively solve the problem of estimating the clutter covariance matrix with the special structural constraints.It is proved theoretically that the covariance matrix obtained by the algorithm can guarantee the above-mentioned non-linear structure.(4)Research on airborne MIMO radar waveform design algorithm based on maximizing mutual information.In this paper,the structured target and clutter covariance matrices of airborne MIMO radar with the transmission waveform are established.Based on this,this paper proposes a waveform design algorithm based on the statistical prior information of target and clutter by maximizing mutual information between target impulse and received signal.The proposed algorithm solves the waveform design problem with the influences of different clutter range bins by the alternation direction method of multipliers(ADMM)algorithm.(5)Research on distributed airborne MIMO radar target detection algorithms.In this paper,the target and the clutter model of distributed airborne MIMO radar system are established.With the aid of the Bayesian detection framework,a point target detector for distributed airborne MIMO radar is proposed in this paper when the prior spectrum structure information of clutter covariance matrix is known.The proposed detector assumes that training data is not available.In addition,a range-spread target detector for airborne MIMO radar without training data is proposed.The detector employs the block sparse Bayesian learning(BSBL)algorithm for estimating the unknown space-time correlated clutter.Based on this,the probability density function of the detection statistics of the proposed range-spread target detector are listed under the two hyphotheses with or without target signal.