Research On Heterogeneous Clutter Suppression For Airborne Radar

To improve the detection performance of remote targets,airborne radar generally operates in a down-looking state.Therefore,the received data not only contain the target information but also include a great number of high-intensity clutter signals.Due to the movement of aircraft,the ground scatterers with different observation angles possess different radial velocities,which makes the clutter spectrum seriously spread in Doppler domain,resulting in the deterioration of traditional one-dimensional adaptive processing.By utilizing multiple spatial and temporal channels for two-dimensional adaptive processing,space-time adaptive processing(STAP)can effectively suppress the space-time coupled clutter,which significantly improves the clutter suppression and target detection performance.Generally,the number of independent and identical distributed(IID)training samples required for STAP should be greater than twice the system degrees of freedom.However,in practical applications,the non-ideal factors such as array geometry configurations,terrain variations,man-made buildings,moving targets on traffic roads,intentional and unintentional electromagnetic interference,will destroy the homogeneous assumption,which makes it difficult to meet the requirement of IID training samples for clutter statistical characteristics estimation.To improve the performance of heterogeneous clutter suppression,some fast and robust STAP algorithms for IID training samples starved cases utilizing the clutter sparsity are studied in this paper,which can be summarized as follows:1.A direct data domain-based fast STAP algorithm is proposed for extremely heterogeneous environments.Existing studies show that the heterogeneity of clutter in distance leads to a wide clutter notch,the discrete interference signals increase the false alarm rate,and the dense moving targets typically result in self-cancelation.Theoretically,these heterogeneity effects can be well mitigated by direct data domain(DDD)-based schemes.However,most of the existing DDD methods suffer from the loss of system degrees of freedom(DOF),the dependence of prior information,and the high computational complexity.To handle these problems,a complex approximate message passing(CAMP)-based fast DDD algorithm is proposed in this paper.The proposed algorithm first utilizes the CAMP framework to quickly estimate the angle-Doppler profile of the cell under test.Then,an adaptive masking matrix designed to avoid target self-cancellation is derived based on the obtained angle-Doppler profile.Finally,the adaptive masking matrix and angle-Doppler profile are both utilized to calculate robust clutter-plus-inference covariance matrix and STAP filter.Compared with the existing algorithms,the proposed algorithm has the following three advantages:(1)The super-resolution ability of CAMP helps the proposed algorithm avoid the performance loss caused by DOF reduction;(2)The designed masking matrix eliminates the moving targets information while retaining the discrete interference,which not only avoids the target self-cancellation effects but also reduces the false alarm detection probability;(3)The proposed algorithm avoids matrix inversion in the process for angle-Doppler profile estimation,which greatly reduces the computational complexity and is conducive to real-time applications.2.A robust STAP algorithm for sparse signal model mismatch is developed.When the sparse signal model constructed by discretized dictionary can accurately represent the clutter data,the sparse recovery-based STAP(SR-STAP)can obtain approximate optimal detection performance by utilizing the super-resolution ability of SR techniques.However,the array gain/phase errors and grid mismatches in practical applications will destroy the accuracy of sparse signal model and deteriorate the performance of SR-STAP algorithms.To handle this problem,we proposed a sparse Bayesian learning-based robust STAP approach.The proposed approach first constructs a novel SR-STAP signal model by exploiting the Kronecker product property of the space-time steering vector.Then,the angle-Doppler profile,array gain/phase errors,and the grid-mismatch parameters are alternatively obtained by utilizing the Bayesian inference and expectation-maximization schemes.Finally,the precise clutter covariance matrix and the corresponding STAP filter are calculated with the above parameters.Simulation results verify that the proposed approach can improve the performance of clutter suppression and moving targets detection in the case of model mismatches.3.A gridless STAP algorithm using prior structure information for CCM estimation in IID training samples starved cases is studied.Considering the limitation of sparse Bayesian learning scheme in the accuracy and efficiency of solving mismatched parameters,a mixed-norm minimization(MNM)based gridless STAP algorithm is proposed in this paper.The proposed algorithm estimates the clutter covariance matrix in continuous domain by solving a semidefinite program(SDP),which is established by utilizing the double Vandermonde structure of space-time steering vector and the compact formulation of MNM.Furthermore,to reduce the computational burden of solving this SDP,we also derive a fast iterative algorithm via the framework of the alternating direction method of multipliers(ADMM).By utilizing the prior information and ADMM framework,the proposed algorithm can eliminate the grid mismatch effect with low computational complexity.Simulation results verify that the proposed algorithm can significantly improve the performance of weak and slow moving targets detection performance.4.A robust gridless STAP algorithm is developed for CCM estimation in heterogeneous environments.Considering the influence of the regularized parameter in MNM-based algorithm,we propose two tuning parameter-free algorithms by utilizing the positive semidefinite(PSD)and block-Toeplitz properties of CCM and the covariance fitting criteria(CFC)from the perspective of clutter statistic property.As regularized parameter-free algorithms,the proposed CFC-based algorithms enjoy a robust property.Furthermore,the corresponding ADMM-based fast implementations are also derived for both CFC-based methods to reduce their computational complexities.Simulation results with both simulated and Mountain Top data demonstrate that high computational efficiency and good performance of proposed algorithms are obtained.5.An elevation adaptive STAP algorithm for non-sidelooking array in the case of range ambiguity is studied.The non-sidelooking geometry configuration of airborne radar will destroy the linear relationship between the normalized Doppler frequency and spatial frequency,resulting in the heterogeneous property of clutter with distance variation.Existing studies show that the heterogeneous property of non-sidelooking clutter is violent in short range and tends to be flat in long range.However,to avoid Doppler overlapping,airborne radar generally works in medium or high frequency pulse repetition frequency(PRF),which will result in range ambiguity and increase the difficulty of clutter suppression for non-sidelooking radar.To handle this problem,a CFC-based elevation adaptive STAP algorithm is proposed.By utilizing the separability of short-range and long-range in elevation domain,an elevation adaptive weight vector calculated by the CFC-based scheme is employed to suppress the heterogeneous short-range to improve the homogeneous property of residual clutter.Then,a traditional STAP is followed to suppress the residual clutter.Simulation results demonstrate the proposed algorithm can greatly improve the slow moving targets detection performance.