Research On 2D-DOA Estimation Of EMVS Array Based On Tensor Decomposition

In the field of signal processing,array signal processing is a significant branch with extensive applications in wireless communication,radar airspace detection,underwater sonar,and other fields.Direction of arrival estimation is one of the main research topics in array signal processing.Traditional radar antennas use scalar sensors,but with the improvement of signal processing requirements,they are gradually unable to meet the detection requirements.In contrast,electromagnetic vector sensors exhibit significant advantages due to their ability to provide two-dimensional direction of arrival estimation and additional polarization information of input signals.The existing relevant research mostly considers the application scenarios of Gaussian white noise,while the actual application of spatial colored noise may be more.In response to this deficiency,this article mainly studies the problem of direction of arrival estimation in colored noise scenarios,and the main achievements are as follows:This article models the output signal of electromagnetic vector sensors and proposes a joint angle estimation algorithm that can be applied to any geometric array of electromagnetic vector sensor systems.The algorithm and core is to obtain Poynting vector by cross product of two special vectors composed of rotation invariant,and then the angle estimation is completed by least square method.The proposed algorithm can provide automatic pairing of two-dimensional angle of arrival,two-dimensional angle of departure,and polarization parameter estimation.In addition,this algorithm is not sensitive to sensor position errors and is superior to various classic algorithms.This article proposes an algorithm for two-dimensional direction of arrival estimation of electromagnetic vector sensors based on tensor decomposition.In order to achieve fast PARAFAC decomposition,this article rearranges the fourth order covariance PARAFAC tensor into a third order PARAFAC tensor,and uses the COMFAC algorithm to estimate the factor matrix.After obtaining the factor matrix,the elevation angle is estimated using the least squares method,and the azimuth angle is estimated using vector cross product.In addition,the polarization state of the source signal can also be obtained through the least squares method,which helps to identify weak signals.This algorithm can be flexibly extended to situations of nonuniform arrays and spatially colored noise.This article designs a numerical simulation to conduct a detailed analysis of the identification performance,complexity,flexibility,and Cramerow bound of the above algorithms,in order to verify their effectiveness.Through numerical simulation results,we can clearly see that the algorithm proposed in this paper has better estimation performance compared to the traditional PARAFAC algorithm in the case of colored noise.It not only has strong suppression of colored noise,but also has higher accuracy of DOA.