Research On Reduced Dimensionality And Real-valued Transformation Method Of Super Resolution Algorithm

Super-resolution algorithm is one of the key technologies for array direction finding,however,in the development stage of the current theory to engineering transformation,the traditional super-resolution direction finding algorithm faces the defects of high computational complexity and strong array structure dependence,which has become the constraint that restricts the direction finding theory to the practical application of military and civilian integration.In recent years,using the characteristics of the signal to improve the performance of the direction finding algorithm has become a research hotspot in this field.Therefore,this paper will use the dimension reduction and real value transformation methods to study the low complexity super-resolution direction finding algorithm,aiming to provide theoretical support for the advancement of direction finding algorithm engineering.The main research contents and theoretical innovations of this paper are as follows:The noise subspace of the beam domain is obtained by using the linear operator method based on the beam space,and the orientation information of the target source is obtained by the polynomial root finding,that is,the BPM-root-MUSIC algorithm.At the same time,the beamforming matrix is decomposed by least squares iteration,and the expression of the reduced-order beam-order root finding algorithm(BPM-B-root-MUSIC)is obtained.Further,in order to avoid root finding within the beam width,a source-order closed rooting algorithm BPM-L-root-MUSIC related only to the number of sources is directly given.Based on the rotation invariance of the non-circular signal,the array output data is realized using a real value transformation.At the same time,in order to avoid large-scale eigenvalue decomposition(EVD)operation caused by conjugate expansion in non-circular signals,whole EVD is replaced by subarray division of covariance matrix and subarray EVD,a new algorithm for quickly extracting noise subspace is proposed(RV-NC-root-MUSIC).The new algorithm adopts the root-seeking method to obtain the source orientation information.At the same time,the polynomial coefficient symmetry relationship is used to greatly reduce the order of the root polynomial,which significantly reduces the computational complexity of the algorithm.Theoretical analysis and simulation experiments show that the two algorithms proposed in this paper implement low-complexity super-resolution direction finding algorithm,and can balance the estimation performance of the parameters with the direction-finding efficiency within a reasonable dynamic range.The research in this paper helps to enrich the theoretical system of the direction finding algorithm.At the same time,it has important practical engineering significance for reducing the research and development cost of various military and civilian direction finding systems and improving the working efficiency of direction finding.