Research On Knowledge-Aided Clutter Suppression

Modern Radars have to deal with various challenges because of the complex and variable clutter environment.The challenges mainly include the clutter suppression for heterogenous and non-Gaussian clutter and huge computational complexity and data storage demands required by radars adopting techniques such as large-scale arrays,signal processing with data from different domains.To solve the above problems,a solution is to exploit the prior knowledges in signal processing,which can effectively reduce the burden caused by computational complexity and huge data storage,thus improving the clutter suppression and target detection performance of radar.In the knowledge-aided signal processing,the clutter covariance matrix(CCM)estimation and clutter suppression performance of adaptive radar filter are two key problems.In previous works,the structured covariance matrix estimation is still not well discussed,also the clutter suppression performance in knowledge-aided signal processing is mainly validated through via simulations.For this reason,the knowledge-aided covariance matrix estimation and the clutter suppression performance of knowledgeaided adaptive radar filter is studied in this dissertation,and an approach to analyze the average signal-to-noise ratio loss(ASCRL)of a radar adaptive filter with parametrically constrained covariance matrix is proposed.The main works are as follows:1.The knowledge-aided adaptive radar signal processing is studied and the prior knowledges of clutter is summarized,including prior knowledges of clutter environment and the structure knowledges of the clutter covariance matrix.The prior knowledges of clutter environment and the corresponding accesses to obtain these knowledges are introduced.The linear model of structured clutter covariance matrix is given.Based on this linear model theory,some typical structured clutter covariance matrices are discussed.2.The knowledge-aided covariance matrix estimation is studied,including estimation using statistical Bayesian theoretics and estimation for structured covariance matrix.Three estimates for PerHermitian structured covariance matrix are then given,and the average signal-to-noise ratio loss of an adaptive filter with PerHermitian structured matrix is derived,which is verified via simulation results.3.The airborne phased array radar target clutter signal model is constructed,based on which the special non-linear structure of summing Kronecker product of multiple spatial and temporal clutter basis matrices with low ranks is analyzed.A Kronecker product permutation operation is introduced,the structured covariance matrix estimation problem of airborne phased array radar is then transformed to a nuclear norm regularized two-norm optimization problem via the permutation operation.The optimal solution of this optimization problem is obtained through the Permuted Thresholded Singular Value Decomposition algorithm.Simulation results show that compared to algorithms where structure knowledges of CCM are not introduced,this algorithm have higher matrix estimation accuracy and better filtering performance in space-time adaptive processing.4.A general approach to evaluate the average signal-to-noise ratio loss of a radar adaptive filter with parametrically constrained covariance matrix is proposed.The constrained Cramer-Rao Bound(CRB)of a CCM estimate is first derived,then then it is used to derive the ASCRL of a radar filter.The constrained CRB of a trace constrained CCM estimate is derived for complex Gaussian as well as compoundGaussian clutter,based on which the ASCRL of the adaptive filter is then derived.Simulation results show that the proposed approach has a good performance.