Researches On Space-time Clutter Spectrum Estimation In SR-STAP

Because the clutter amplitude is larger than the target amplitude in the actual scene,the conventional method is difficult to eliminate the strong clutter under the configuration of the phased array antenna.The STAP is an efficient tool for radar signal processing.It can make full use of the information of time domain and space domain to eliminate the strong clutter in the background,and then to acquire high target detection performance.Then,the STAP is a key technique in radar signal processing and becomes a research hotspot in AWACS.However,the STAP needs much number of IID samples to estimate the clutter covariance matrix(CCM),which is difficult to be acquired because the clutter in actual scene is nonstationary.Thus,it will lead to insufficience of clutter cancellation,In addition,the computational complexity of the coefficient of STAP filter in the process of solving the CCM inverse is three order.Such computational complexity will seriously affect the clutter suppression of the STAP or the real-time detection of the target.In order to reduce the sample quantity and the computational burden,some scholars study the suboptimal STAP technique in-depth.But the suboptimal STAP technique is still difficult to be applied in actual application because the sample quantity required in the suboptimal STAP is still too much.In addition,the suboptimal STAP method reduces only the computational burden,but not the computational complexity.With the development of the sparse recovery(SR)technique,some scholars take the sparse recovery into the STAP,and propose a sparse recovery based STAP(SR-STAP).The SR-STAP makes use of the sparsity of the scatters on the spatial-Doppler plane,and can estimate the CCM by using very little amount of sample or the data on the test cell.Basis mismatch,multiple measurement vector problem,conformal Array,and direct data are research focuses of the SR-STAP in the recent years.The estimation of the clutter space-time spectrum is the key in the SR-STAP.The accuracy level of the estimation relates to the performance of the SR-STAP.In addition,the high complexity of computation of the STAP coefficient can still influence the performance of the SR-STAP,which is the outcome of the time efficiency of the estimation.Then,the content of this paper is on the fast and accurate estimation of the space-time spectrum.Since clutter and target are overlapped on the spatial-Doppler plane for the wideband radar configuration,which will lead to the inaccuracy of the estimation of the space-time clutter spectrum.Then,it needs to build a wideband configuration based SR-STAP.Thus,the following is our works:1.A fast method for estimating the space-time spectrum is proposed.According to the relevant literatures,the focal underdetermined system solver(FOCUSS)is an efficient algorithm to estimate clutter space-time spectrum in SR-STAP.However,its computational complexity is very high,and then it is not appropriate to be employed in actual clutter suppression in SR-STAP.Then,we propose a fast FOCUSS algorithm based on the bi-conjugate gradient(BICG),i.e.,BICG-FOCUSS.In this algorithm,the section with large of computational burden in the FOCUSS,i.e.,the Moore Penrose inversion,is replaced by a solution of a linear Hermitte equation.The BICG method can solve a linear equation with lower computational complexity,and we employ the BICG to solve this linear equation.Simulations show that the BICG-FOCUSS can faster converge to a sparse solution without loss of accuracy,and we employ the BICG-FOCUSS in radar system to estimate the clutter space-time spectrum.Experiments show that the BICG-FOCUSS can faster and accurately estimate the clutter space-time spectrum in SR-STAP.STAP filter is obtained by the multiplication between the inversion of the clutter covariance matrix and the target space-time steering vector.In the paper,we regard the coefficient as the solution of the linear equation and employ the BICG to solve this linear equation.Since the computational complexity of the BICG is two orders of magnitude,this method can obtain the STAP filter coefficient in a very short time.2.An algorithm which can eliminate the basis mismatch of the space-time spectrum is proposed.In this paper,we propose a parameter-searched orthogonal matching pursuit(PSOMP)method to eliminate the basis mismatch appeared in SR-STAP.This method takes the frequency errors between the scatter and its nearest atom in the matrix as unknowns,and uses an optimization method to compensate the frequency errors.In the PSOMP,the selected atom in the OMP is carried two parameters and then the optimal parameters are searched by solving an optimization so that the adjusted atom in the basis matrix can match with the true scatters in the data.It can be derived that the optimization problem is approximately equal to a convex optimization,and then the steepest descent method or the Newton method can be employed to search the optimal parameters.The gradient vector and the Hesse matrix can be given as an explicit formula.Compared to the OMP method,the extra computational burden of the PSOMP is produced by solving the optimization problem.In the end,the PSOMP algorithm is compared with the existing CSDL algorithm and POMP algorithm.The results show that the PSOMP algorithm can effectively compensate the mismatch ambiguity,and the compensation can be fulfilled in a very short time.3.The research of the STAP is expanded.We propose a STAP model for wideband radar.We take the linear frequency modulated(LFM)signal as the transmitting signal in the radar,and obtain the signal form after match filter for the signal returned from a scatter.We make space sampling and coherent pulse sampling of the signal form,and define Doppler frequency,Doppler modulating frequency,spatial frequency,spatial modulating frequency,and Doppler-spatial cross frequency.It is known that these frequencies are determined by the spatial frequency and the Doppler frequency.Then,we define a spatial-frequency steering vector,and then know that the echo in a range cell is a linear combination of such steering vectors.This STAP model is called as LFM-STAP.We derive the clutter trajectory under the side looking mode and the non-side looking mode.In addition,because of shortage of IID samples,we discretize the spatial-Doppler frequency plane and employ the SR technique into LFM-STAP to estimate the clutter space-time spectrum.Experimental results show that our proposed model can efficiently estimate the clutter space-time spectrum in wideband configuration when the LFM signal as the basic signal in STAP.