Research On DOA Estimation Of Distributed Sources Based On Coprime Array

Scattering,reflection and diffraction in a complex environment will cause the signal source to expand at a certain angle in space.In this case,the target signal source needs to be established as a distributed source model.At present,the direction of arrival(DOA)estimation algorithms for distributed source models are mostly based on the uniform arrays,and the application of sparse coprime array in the field of distributed source DOA estimation faces applicability problems.Most of the existing algorithms are based on the assumption that the distributed source signals are Gaussian signals and circular signals,and seldom consider the characteristics of non-Gaussian signals and non-circular signals.Compared with uniform array,sparse coprime array is widely used in DOA estimation of point source models due to its advantages of large array aperture and small mutual coupling error.Moreover,it is of great significance to improve DOA estimation capability of distributed source by using special characteristics.Therefore,this paper considers to solve the applicability of the coprime array in the field of DOA estimation for distributed source model,and proposes several high-precision and high-degree of freedom distributed source DOA estimation algorithms.The main innovation are as follows:1.Existing algorithms are limited by the disadvantages of small aperture and large mutual coupling error of uniform array.To solve this problem,a distributed source angle parameter estimation algorithm based on the common spectral peak search in the coprime array is proposed which solves the applicability of coprime array in the condition of distributed source model.Based on the single source incident scene of general distributed sources,the fuzziness mechanism is derived,and the intersection of DOA estimation results of two sparse uniform subarrays is proved to be the true value.Based on this,the angle fuzziness is eliminated.The algorithm is no longer limited to the structure characteristics of the original sparse coprime array,which makes the algorithm have strong applicability in complex application scenarios.The experiments show that this algorithm has higher-performance than the existing algorithm with the same array number.2.Most existing algorithms are based on the assumption that the signal is a Gaussian signal.When the signal is a non-Gaussian signal,the second-order statistics can't fully describe the signal characteristics,so a certain amount of information is lost,leading to the limited performance of DOA estimation.To solve this problem,an estimation algorithm of distributed source angle parameters based on the fourth-order cumulant is proposed,taking both the non-Gaussian high-order cumulant and the virtual array property of the mutual-matrix into account.Different from second order statistics,when the signal is non-Gaussian signal,fourth-order cumulant contains more useful information.The proposed algorithm firstly establish the received signal model of coprime array,select the appropriate fourth-order cumulant form,thus derive the mechanism in extending the virtual array by fourth-order cumulant in distributed source model,expand the array aperture,improve the estimation degree of freedom,solve the applicability problem of fourth order cumulant and coprime array in the distributed source model;Secondly,the fourth-order cumulant is de-redundant and treated as the covariance matrix of the received signal of the virtual array.Finally,the center azimuth and angle expansion parameters are solved by two-dimensional spectral peak search.Simulation results and analysis show that the proposed algorithm are better than the existing algorithm with the same physical array sensors number.3.In the existing algorithm,when the circular signals are mixed with the non-circular signals,the circular signals cannot be used to extend the virtual array,which will limit the improvement of the aperture and degree of freedom of the array.Moreover,the distributed circular signal and distributed non-circular signal in the same direction cannot be distinguished.To solve this problem,an algorithm based on the separation of circle and non-circle signals is proposed.The proposed algorithm firstly studies the received signal model in the mixed scene of distributed circular signal and distributed non-circular signal in the coprime array.Then the distributed non-circular signal and distributed circular signal are solved by DOA estimation.For non-circular signals,the virtual array is obtained by using the elliptic covariance matrix vecturization method,which makes the received signals into the equivalent single beat form.The equivalent covariance matrix is obtained by spatial smoothing processing,and then the center azimuth and angle expansion parameters of non-circular signals are solved by using spectral peak search algorithm or the root algorithm with lower complexity.For circular signals,the estimation results of distributed non-circular signals are substituted in reverse,and the covariance matrix of non-circular signals is reconstructed to obtain the equivalent covariance matrix containing only distributed circular signals.Similarly,the spectral peak search algorithm or the root method is used to solve the center azimuth and angle expansion parameters of distributed circular signals.Theoretical analysis and simulation results show that the proposed algorithm are better than the comparison algorithm through vectorization and signal separation and reconstruction,so it can achieve underdetermined estimation and obtain higher parameter estimation accuracy.