Study On DOA Estimation Algorithm Based On Incomplete Vector-Sensor Array

The electromagnetic vector-sensor is composed of various polarized antennas with different polarization characters.The polarization sensitive array(PSA)which consists of electromagnetic vector-sensor can obtain the amplitude,phase,waveform and polarization information.The complete electromagnetic vector-sensor can obtain spatial,time and six-domains polarization information,compared with the traditional scalar array,PSA has the advantages of better anti-jamming and higher resolution ability.An incomplete electromagnetic vector sensor contains subantennas of the complete electromagnetic vector sensor,and also keep the advantage with fewer receiving data,less mutual coupling and volume of space.Therefore,the incomplete electromagnetic vector sensor has been widely used in the field of military and civilian,such as radar,sognar,navigation and communication systems.This dissertation focuses on the mathematical model,angle resolving capability and parameters estimation algorithm for the incomplete electromagnetic vector sensor,to improve parameters estimation accuracy and reduce the complexity.The three contributions included in this disseration are described as follows:The first part studies the problem of resolution of the distributed array which is composed of single-dipole antenna,and we propose a rank-deficient MUSIC algorithm.First,mathematical model is built for the incoming signal response of the arbitrary direction single-dipole antenna,and the definition of resolution is the capacity of distinguishing two closed signals in space.By studying the relationship between the resolving capability and similarity of steering vector,and combined with the source estimatin and rank-defect MUSIC algorithm,it is concluded that the distributed array has a higher resolving capability than the traditional scalar array.Simulation results show that PSA have a better angular resolution than the scalar array.The second part studies muktidimensional parameters estimation algorithms for the uniform linear array which is composed of dual-cross dipoles.First,using quaternion model for reconstructing the receiving data,and then the receiving data dimensionality is reduced by half.When signals are circular,the reconstructed data covariance matrix can be further reduced.Then using the Root-MUSIC algorithm to estimate the DOA and the generalized eigenvalue to estimate the polarization angle.There is no need to do a spectral peak searching all the way while improving the estimation accuracy and reducing the computational complexity significantly.However,the above algorithm is established under the premise that the incident signal is uncorrelated,and then we focus on the problem of mixed signals estimation and propose a parameter estimation algorithm based on quaterion model.Proposed algorithm under the quaterion model use Toepltiz matrix to separate the independent and coherent signals.Root-MUSIC algorithm is used for estimating the DOA of independent signals,and then the reconstructed characteristic vector is used for estimating the DOA of coherent signals.This algorithm makes the independent signals separated from mixed signals,and extend array aperture with high accuracy and low complexity.Simulation results vertify the favorable performance of the two proposed algorithms.The third part studies a problem of array extension based on the three-cross dipoles.First,we built a mathematical model of the co-prime array which is composed of three-cross dipoles,and make the co-prime array more reasonable.Then a MUSIC algorithm based on a the reconstructed matrix is proposed for the co-prime array,this algorithm makes full ues of all information of the three-cross dipoles to form the virtual array which has more array elements.After the virtual array is formed,turning the uncorrelated signals into coherent signals,therefore the covariance matrix of the virtual receiving data need to be untied suing the forward and backard spatial smoothing technique method.This proposed algorithm can estimate the number of signals which is more than the number of array element.However,the performance of this proposed algorithm fell sharply or even failure at a low snapshot number.Aiming at these problems,we propose a sparse algorithm based on the coprime array.The proposed algorithm make the covariance matrix sparse based on the compressed sensing theory,and then use the characteristics of PSA to remove noise of the covariance matrix.Finally,this new covariance is used to estimate DOA using Basis Pursuit Denoising method.The proposed algorithm can still estimate DOA at a low snapshot number with unknown source number.For the problem of the existed pseudo peak in the power spectrum,we designed a sliding window menthod to find the real peak.Simulation results show that both algorithms can achieve array sctension.The MUSIC algorithm based on the reconstructed matrix has a better performance at a high snapshot number.The performance of the sparse algorithm based on the reconstructed matrix at low snapshot number is better than the former,and the sparse algorithm does not require the estimation of source number.