DOA Estimation Algorithms Based On Subspace Approach

Direction of arrival(DOA) estimation focuses on the research of the estimation ability of the processing system composed by multiple sensor arrays. The purpose of DOA estimation is to confirm the spatial positions of signals which we are interested in, in other word, it is to estimate the signal angles arrived at the array’s reference element. Theoretically, the research of DOA estimation algorithm can greatly improve the estimation accuracy of the space angles within the bandwidth of the system and reduce the run time of the signal processing.In this thesis, we focus on the research of DOA estimation algorithms based on subspace analysis, aim to improve the estimation accuracy and reduce the computational complexity of the algorithms. Some details are as follows:1. The sensor array model is built and the corresponding mathematical model is presented. Based on the mathematical model, some necessary theoretical knowledge is provided as well.2. Some commonly used subspace algorithms, including of the principles and improve-ments of MUSIC and ESPRIT, are introduced, and then their performances are analyzed and compared using MATLAB. At the same time, a fast subspace algorithm-PM algo-rithm is described, as the theoretical basis of the follow-up study.3. On the DOA estimation of coherent signals, firstly, the traditional spatial smoothing algorithms are analyzed. Secondly, against on the problem of the spatial smoothing algorithms which can not use the array aperture effectively, an improved matrix re-construction algorithm is proposed. In the proposed algorithm, a full rank Hermitian Toeplitz matrix through rearranging the correlation matrix of the array reception data is constructed. In order to improve the estimation accuracy and reduce the computational complexity of the algorithm, the method of conjugate rearrangement and real value are used as well. The MATLAB simulation results show that the proposed matrix recon-struction algorithm has excellent performance. 4. On the2-D DOA estimation, the ESPRIT algorithm and the PM algorithm based on the L-shaped array model are derived. And then, the two algorithms are compared with those based on the double parallel array model using MATLAB. The simulation results show that the ESPRIT algorithm based on the L-shaped array model has excellent per-formance, while the result of the PM algorithm is unsatisfactory.