Direction Of Arrival And Polarization Estimation Based On Constrained Tensor Decompositions

The estimation of DO A and polarization parameters is one of the important applications in the field of array signal processing. It was widely used in many fields such as communication, radar, sonar. Compared with the traditional scalar sensors, electromagnetic vector sensor array can sense all the three-dimensional information of the space-time-polarization, which has advantage in the use of information. Therefore it can achieve a better accuracy of DO A and polarization joint estimation.Because the output signal of the electromagnetic vector sensor array has a space-time-polarization three-dimensional structure. The traditional method with matrix factorization needs to convert the original signal into a matrix to process, but this process will destroy the three-dimensional structure of the original signal. So the high-dimensional processing tools based on tensor have attracted increasing attention. Because of the characteristics of the signal and the location of the arrays, the model will have some priori information, such as Uniform Line Array (Vandermonde structure), constant modulus signal (constant modulus constraint), no-correlation between the signals (orthogonal constraint), signal statistics independent (ICA), so the constraint tensor decomposition method can be used to estimate DOA and polarization parameters of the signal and calculate the space and polarization steering vector matrix associated with them. The constraint tensor decomposition algorithm utilizes not only the three-dimensional structure of the original signal but also the priori information of the signal and the array. This thesis mainly studies DOA and polarization parameters joint estimation of the electromagnetic vector sensor array based on constraint tensor decomposition, the main achievements include:● Studying DOA and polarization parameters estimation of electromagnetic vector sensor array based on waveform constrained, we proposed a fifth-order symmetric tensor decomposition algorithm with combination of ICA:First, we construct the fifth-order symmetric tensor by calculating the fourth-order cumulant of the incoming signal. Then we use this decomposition algorithm to estimate the polarization steering vector matrix. At last DOA and polarization parameters of the signal can be estimated from the polarization steering vector. We proposed a third-order tensor decomposition algorithm based on constant modulus and orthogonal constrained:First we use this decomposition algorithm to estimate the polarization steering vector matrix to estimate DOA and polarization parameters of the signal. Simulation results show that the tensor decomposition algorithms based on waveform constraint can improve the estimated accuracy of DOA and polarization parameters.● Studying DOA and polarization parameters estimation of electromagnetic vector sensor array based on waveform and array constraint, we propose three algorithms:tensor decomposition algorithm based on Vandermonde constant modulus constraint、based on Vandermonde and orthogonal constraint、based on Vandermonde constant modulus and orthogonal constraint.These three algorithms are all to estimate DOA and polarization parameters of the signal from the polarization steering vector. Simulation results show that the tensor decomposition algorithm based on orthogonal and Vandermonde structure can obtain the best estimated accuracy. It indicates that the constraint combined with the waveform and array shape is beneficial to the DOA and polarization parameters estimation. But the algorithm combined with three constraints can’t get great results, which indicates an increasing number of constraints won’t improve the estimated accuracy.