Estimation Of Signal Parameters Polarized Array Based On Sparse Recovery

Source Parameter Estimation has important significance of research and value of application involving the wireless communications, radar, sonar, etc. The existing source parameter estimation algorithms are mostly theory of subspace-based, they utilize the orthogonality of signal subspace and noise subspace to realize the source parameters estimation, their performance depend on that high Signal to Noise Ratio, the more snapshots number and the information of sources exact number. In recent years, the theory of sparse signal reconstruction is applied to the field of source parameter estimation, the studies show that a class algorithm of sparse signal reconstruction source parameter estimation can overcome the limitations of subspace-based methods in terms of resolution, robustness and priori conditions, but existing class of sparse signal reconstruction source parameter estimation methods have deficiencies as the following three aspects:(1) the receiving array commonly are scalar sensor array, the information polarization of the source has not been fully explored;(2) the process of reconstruction is often inequitable constraint, it has influenced the precision of the source parameters and the estimated performance;(3) the mesh of multi-dimensional results in higher computational complexity when the estimation of multi-source parameters such as azimuth and distance. Therefore, it is necessary that researching indepth the source parameter estimation of polarization sensitive array based on sparse signal reconstruction sequentially, and improving the theory system of parameter estimation for sparse signal reconstruction.The main contributions and innovation points of this paper as follow:(1) Aiming at the unfair constraints, the character of capon spectral function which high accuracy of approximations 0 norm and strong sparsity, the new algorithm is proposed that the mixing near and far field to the constraint of capon spectral function based on fourth-order cumulant. The algorithm sparse indicates the fourth-order cumulant which contains only angle information and obtains angle estimation through the reconstruction, on this basis, the angle estimated value is substituted into the overcomplete sparse mixing matrix to obtain the range estimation. Compared with the existing class methods of sparse signal reconstruction with mixing near and far fields, the proposed algorithm can effectively improve the accuracy of estimation.(2) Utilizing the performance of non-convex function which approximates better 0 norm, it is proposed the algorithm of DOA estimation with non-convex penalty functions based on sparse signal reconstruction, we solve the problem of unfair constraint in processing of reconstruction further. Extending the L-shaped array, the novelty algorithm is proposed that low complexity of joint estimation of elevation and azimuth. The special property of L-shaped array to divide the submatrix, so that the direction matrix of first sub-array contains only the elevation angle information, the direction matrix of second sub-array contains both information of elevation and azimuth, we use non-convex penalty function and theory of DC decomposition to obtain the elevation estimation, the elevation estimation is substituted into sparse observation of second sub-arrays related to obtain azimuth estimates. Compared with existing algorithms of 2D-DOA estimation, the proposed algorithm has higher accuracy of estimation, by solving for the source multi-parameter estimation step-step, it effectively solves the problem of high complexity of the sparse signal reconstruction for the source multi-parameter Joint Estimation.(3) With polarization sensitive array of COLD, the combination of sparse signal reconstruction and quaternion, the new algorithm is proposed that estimation of azimuth and polarization. The algorithm structures the receiving model of quaternionpolarization based on the theory of quaternion, we utilize the orthogonal of quaternions to separate azimuth and polarization, through non-convex constraint penalty function to reconstruct DOA estimation, and achieve polarization parameters estimation(polarization angle amplitude and phase) through the search peaking. And compared with the existing algorithms of polarization sensitive array source parameter estimation, the proposed algorithm avoids the multidimensional mesh of multi-sources estimation sparse signal reconstruction, while improving the accuracy of source parameter estimation, the resolution of sources and noise robustnessto noise.This four novelty algorithms are proposed in the sparse signal theoretical framework around the factors such as mixing far and near fields, polarization sensitive array, constraint functions, it provides reference of multi-parameter estimation methods based on source sparse signal reconstruction under polarization sensitive array for further study.