Research On DOA Estimation Of Non-Uniform Sparse Arrays

Direction of Arrival(DOA)estimation is the main research content of array signal processing,and it is widely used in radar,sonar,electronic countermeasures,radio astronomy,speech signals and other fields.Traditional arrays are limited by the number of physical array elements when DOA estimation is carried out,and the number of estimable signals cannot exceed the number of array elements.In order to solve the problem of underdetermined DOA estimation in which the number of signals exceeds the number of elements,non-uniform sparse arrays attract people's attention because of their special array structure.Nested array and coprime array are two kinds of classical nonuniform sparse arrays.They can improve the estimation performance of DOA and realize underdetermined estimation by using the generated difference coarrays.However,there are still many problems to be solved in the study of non-uniform sparse array.In this paper,we focuses on the one-dimensional and two-dimensional DOA estimation of non-uniform sparse arrays.The main work and innovation are as follows:(1)A new type of non-uniform sparse array named sum and difference nested array is proposed.VCAM algorithm can be used to generate corresponding sum and difference coarray for sum and difference nested array.Compared with traditional nested array and coprime array,sum and difference nested array can provide more degrees of freedom and larger array aperture based on the concept of sum and difference coarray.Therefore,the estimation performance of DOA can be effectively improved.(2)A high precision pair-matching algorithm based on L-shaped nested array is proposed.The one-dimensional angle information of the signal is obtained by using the difference coarray of the subarray on each axis,and the estimation of respective array manifold can be obtained.In a theory,there must be a permutation matrix which can make the angles estimated by each axis matched one by one.The array manifold,permutation matrix and autocorrelation matrix of received signals can be combined to fit the crosscorrelation matrix of received signals of two axes,thus establishing and solving the optimization problem,so as to realize pair-matching and complete two-dimensional underdetermined DOA estimation.The algorithm has high detection probability and eliminates the infuence of noise,which improves the performance of DOA estimation.(3)Two new types of non-uniform sparse parallel arrays are proposed,namely two parallel nested array and symmetric parallel nested array.Based on the special array structure of two parallel nested arrays,the concept of difference coarray is extended to two dimensions,and the two parallel difference coarray is generated.Similarly,parallel difference coarray can be generated based on symmetric parallel nested array and by matrix completion method.Both the two parallel difference coarray and the parallel difference coarray are two-dimensional virtual arrays,and the number of array elements is far more than the physical array itself.Therefore,the accuracy and performance of twodimensional DOA estimation are improved.