Joint Direction Of Arrival And Polarization Estimation Via Canonical Polyadic Decomposition With Constant Modulus Constraints

The research of electromagnetic vector-sensor(EMVS)array has always been an enduring topic in the field of signal processing.It was widely used in many fields such as radar,communication and other fields.Compared with the traditional scalar sensors,EMVSs can preceive space-time-polarization three-dimensional information,that is,the received signal is a third-order tensor,and its information is more comprehensive.Therefore,the EMVSs array can achieve a better accuracy of DOA and polarization joint estimation,and it is widely used.The traditional methods of array signal processing are mostly based on matrix decomposition,which can destroy the three-dimensional structure of array output.Based on this reason,the high-dimensional processing tools based on tensor have been introduced into array signal processing,which can intuitively describe the third-order tensor model of EMVSs array output.In practical application,the incident signal usually has some prior information,such as constant modulus constraint,van der Mond constraint,orthogonal constraint,etc.So we can use the constrained tensor decomposition method to estimate the sources signal matrix,space-domain steering matrix and polarization-domain steering matrix,and further estimate DOA and polarization parameters of the signal from the space-domain steering matrix and polarization-domain steering matrix.This paper mainly studies joint DOA and polarization estimation via canonical polyadic pecomposition with constant modulus constraints.The main results are as follows:We study the joint estimation of DOA and polarization parameters of EMVSs array based on full constant modulus constraint,and present a CPD algorithm based on full onstant modulus constraint: First,we use the classical blind source separation algorithm to recover the sources signal.Then,we obtain the space-domain steering matrix and polarization-domain steering matrix by rank-1 approximation.In addition,we propose an algorithm for estimating DOA and polarization parameters based on cross product.It estimates DOA and polarization parameters from polarization-domain steering matrix and improves the accuracy by using space-domain steering matrix.Finally,we analyze the identification conditions of the proposed algorithm and the Cramero bounds under the specific conditions.The simulation results show that the proposed algorithm is superior to CPD and other iterative algorithms in identification and estimation accuracy.We study the joint estimation of DOA and polarization parameters of EMVSs array based on partial constant modulus constraint,and present a CPD algorithm based on partial onstant modulus constraint: First,we use the classical blind source separation algorithm to estimate the constant modulus information,including it's space-domain steering matrix and polarization-domain steering matrix.Then,we can estimate the non-constant modulus information via CPD.Final,we can estimate the DOA and polarization parameters via cross product algorithm.In addition,we analyze the identification conditions of the proposed algorithm.The simulation results show that the proposed algorithm is superior to CPD and other iterative algorithms in identification and estimation accuracy.