High-resolution Parameter Estimation Algorithm For The Electromagnetic Vector Sensor Array

The electromagnetic vector sensor(EMVS) array is a new type of array, which can obtain the electromagnetic wave polarization information. Compared with scalar array,EMVS array has the advantages of better anti-jamming and higher resolution ability.The EMVS array signals are the multidimensional signals which are included in spatial and polarization domain. The direction of arrival(DOA)-polarization parameter estimations and the array error calibrations based on the EMVS array are very important.This dissertation focuses on the high resolution algorithms for the sparse EMVS array.The key innovations in this thesis are as follows:1. The calibration and compensation of orientation error, amplitude and phase error,coupling error with EMVS array is proposed in this dissertation. By using a single calibration source without exact knowledge of its DOA and rotating the transmitting antenna’s 90 degrees around z axis and x axis, and the array with misorientation receives the three signals whose DOAs have special relationship. According to the subspace theory and the relationship of Poynting vectors, the calibration source’s DOA and the array misoriention are precisely estimated. To calibrate the amplitude and phase errors, a reference signal with unknown DOA is utilized in this dissertation. Two sets of data collected by the dipole triad before/after rotation around its x-axis are used. The closed-form solution of DOA, amplitude and phase errors are got by making use of the electromagnetic field ratio and DOA before/after rotation. Based on the theory of electromagnetics, the mutual coupling matrix model among the elements of an electromagnetic vector sensor is put forward. Using one reference signal source and according the subspace theory, the coupled array manifold is obtained. According the relation between the coupled and the uncoupled array, the least squares solution of normalized coupling matrix is obtained by matrix operations. The whole array coupling error is calibrated by array elements being calibrated one by one.2. The high resolution parameter estimation algorithms based on sparse EMVS array are studied. The larger the aperture is, the higher the DOA estimated accuracy becomes, accordingly the higher the angle resolution turns into. The dissertation mainly includes four parts. The first part is described as the DOA and polarization parameter estimation based on the uniform concentric circular array. The phase differences between two array elements are acquired from the steering vectors of inner circle, it canbe used to give rough but unambiguous estimates of DOA, they are used as coarse references to disambiguate the cyclic phase ambiguities on outer ring circle. The second part is described as the DOA and polarization parameter estimation based on the sparse uniform concentric semi-circular array(UCSA) consisting of concentred orthogonal loop and dipole(COLD) pairs. The actual array steering vector can be transformed into a virtual one without additional computation. By applying virtual transformation operation to spatial steering vectors, two new spatial steering vectors of arrays, whose inter-element spacing are less and much larger than half a wavelength respectively, are obtained, the cyclic phase ambiguities are disambiguated effectively. These estimates are automatically matched and the parameter estimation accuracy is increased considerably. The third part is about the two-dimensional DOA ambiguity resolution algorithm, which is applied to the high frequency eletromagnetic wave interferometer.By means of implementing one or more times virtual transformation operation on phase differences between the original circular array elements and the reference array element,the unambiguous phase differences between the virtual array element and reference array element are achieved. The unambiguous phase differences are used to give rough but unambiguous estimates of DOA, which are used as coarse references to disambiguate the cyclic phase ambiguities in phase differences between the original array element and the reference array element. The last novel algorithm is described as follows. The non-uniform L-shaped spatially spread loop and dipole(SSLD) array whose inter-element spacing is greater than half a wavelength is studied. The direct sampling and the corresponding delayed sampling data are used to construct the data correlation matrix. According to the relationship of array manifold vector between electric dipoles and magnetic loops, the polarization parameters are estimated. The unambiguous phase estimates are acquired by applying virtual baseline array transformation to the spatial steering vectors, they are used as coarse references to disambiguate the phase between two adjacent array elements on the array, the high accuracy DOA estimates is obtained.3. Two quaternion-ESPRIT-based algorithms for EMVS array to estimate the DOA and polarization are proposed. In the first algorithm, a joint estimation quaternion-ESPRIT algorithm of frequency, DOA and polarization based on the L-shaped array is proposed.Two sets of synchronous sampling data are used to construct the data covariance matrix.With the subspace theory and the quaternion eigen-decomposition of data covariancematrix, the array steering vector estimates of the whole array are obtained. The direction cosine of the x-axis and y-axis are acquired by applying block operation to the spatial steering vectors, hence the two-dimensional DOA estimates are got. According to the array steering vector matrices, the dipole sub-array steering vectors of the x-axis and y-axis are reconstructed, their relationships are used to get the estimates of polarization parameters. In the second algorithm, the cylindrical conformal array model is utilized,the procedure is described as follows. Firstly, the estimations of the array steering vector and the elevation are obtained by the quaternion eigenvalue decomposition of array data covariance matrix. Secondly, the azimuth estimations are got according to the relationship between direction cosine and spatial steering vector. Finally, the polarization parameter estimations are obtained by using the relationship between the dipole and loop sub-array steering vector, which are constructed by a real and three imaginary part of the array steering vectors. The closed-form solutions of signal parameter estimates are given and parameters can be automatically matching. Compared with long vector method, the quaternion method has the advantages of lower computation complexity and higher estimated accuracy.