Research On 2D DOA Estimation Algorithm For Array Signal

Two-dimension(2D) direction-of-arrival(DOA) estimation of multiple incident signals has important use in radar, sonar, communication and biomedical engineering. A most famous 2D DOA estimation algorithm based on subspace is super-resolution. This method estimates the DOAs of the sources by using the space-time structural features of the sources. The demand for the computational cost, estimation accuracy and angular resolution of DOA estimation algorithm become more and more high as the 2D DOA estimation is applicable for engineering practice. We conduct research on 2D DOA estimation algorithm to enhance these three indices of the DOA estimation algorithm in this dissertation, and make some meaningful research results. The main research work of this dissertation is summarized as follows:Firstly, the correlation matrix is computationally intensive and performing singular value decomposition(SVD) or eigenvalue decomposition(EVD) on the correlation matrix is time-consuming in engineering practice. A computationally efficient subspace algorithm is proposed for 2D DOA estimation with L-shaped array to solve the problem in this dissertation. The algorithm requires neither calculating the correlation matrix nor performing SVD or EVD on the correlation matrix between the observed vectors. The problem is solved by dealing with three vectors composed of the first column, the first row and diagonal entries of the correlation matrix. The method has the advantages of high DOA estimation accuracy and easy pair-matching. To fully utilize the correlation matrix a modified method is proposed, and the modified method can further improve the DOA estimation accuracy and pair the 2D angles of the sources automatically.Secondly, the larger the array aperture, the higher the DOA estimation accuracy.This dissertation proposes two array aperture extension algorithms. One is called ”Array aperture extension algorithm with T-shaped array”, the other is called ”Array aperture extension algorithm with L-shaped array”. The first algorithm enlarges the dimension of correlation matrix by utilizing the conjugate symmetry of T-shaped array together with the property that the signal covariance matrix is real diagonal matrix,and the second algorithm enlarges the dimension of correlation matrix by using the rotational invariance in conjunction with the property that the signal covariance matrix is real diagonal matrix. It is equivalent to extend the array aperture. By this treatment, the DOA estimation accuracy is improved.Thirdly, we derive the orthogonality of the transition matrix from the estimated signal subspace to the array manifold matrix, and propose a pair-matching algorithm for 2D DOA estimation according to the property. The algorithm can be applicable to planar array consisting of two uniform linear array, such as, V-shaped array, L-shaped array,T-shaped array and cross-shaped arrays. In addition, we propose a DOA estimation algorithm for coherent signals in the presence of uncorrelated and coherent signals by employing the orthogonality in one-dimension DOA estimation. Since the algorithm uses more information to estimate the DOA of coherent signals the DOA estimation accuracy is improved.Fourthly, the subspace algorithms are super-resolution, however, the performance of subspace algorithms will decline rapidly for closely spaced sources. A 2D DOA estimation algorithm is proposed for closely spaced sources in this dissertation. We formulate two special matrices by using matrix multiplication. The nonzero eigenvalues and corresponding eigenvalues contain the pair-matching and DOA information respectively.The DOAs of the sources are estimated by utilizing the their eigenvectors, and the DOA estimation accuracy is high even if the incoming signal is close to each other. The pairmatching procedure is carried out by utilizing their eigenvalues, and the pair-matching procedure does not result in the pair-matching failure. In addition, we also discuss the reason that the presented method can distinguish closely spaced sources.Finally, the 2D DOA estimation algorithm with automatic pairing is also an important research. This dissertation proposes two 2D DOA estimation algorithms with automatic pairing. In the first algorithm, we get the signal subspace by rearranging the elements of the correlation matrix in the first algorithm. The proposed method does not need to perform SVD and EVD of the correlation matrix. Thus, the computational burden of it is low. In the second algorithm, we first partition the correlation matrix into two sub-matrices, and then we multiply one sub-matrix by the Moore-Penrose inverse of the other sub-matrix to get a new matrix. The eigenvectors and eigenvalues of which contain the information of azimuth and elevation angles respectively, and the corresponding relationship between the eigenvalues and eigenvectors can be used to pair 2D angles. The method has the features of high accuracy and low computational cost.