空间谱超分辨算法研究-信息与通信工程专业毕业论文.docxVIP

哈尔滨工业大学工学硕士学位论文 哈尔滨工业大学工学硕士学位论文 II II Abstract The spatial spectrum estimation is a spatial signal processing technology founded in the 60s, 20th century. It can also be considered as a new technology developed on the beam forming technology, zero point technology, as well as the time domain spectrum analysis technology. The main target is to study all the algorithms about increasing the accuracy, the angle resolution, and the operation speed in estimating the angles of the spatial signals within bandwidth. From the development of the spatial spectrum estimation theory, an apparent trace of the main algorithms could be found. The algorithms develop from the linear prediction algorithm to the subspace decomposition algorithms, until now to the subspace fitting algorithms. The development of spatial spectrum estimation algorithms now mainly lies on the fastness and the estimation accuracy. Some new algorithms like single snapshot algorithms and the algorithms without eigenvalues decomposition are proposed for this purpose. The thesis focuses on fast improvement of the classic spatial spectrum estimation algorithms. The performance in source estimation and the calculation are the top judge norm in this thesis, and algorithms are compared searching for the possibility in fast improvement among the spatial spectrum algorithms. First, some of the classic spatial spectrum algorithms are compared in theory, and the calculation and performance of those algorithms are taken out as the norm. Then, take the multiple signal classification algorithm (MUSIC) as the main target, through theory analysis and simulation, the few factors that may influence the performance of the spatial spectrum algorithms, like small signal snapshots, near or coherent sources, spatial colored noise, and array errors, etc, are determined here. Fast DOA estimation algorithm using pseudo covariance matrix is the algorithm used in thesis, it could gain a good estimated accuracy of the directions of source