应用于任意阵列结构的自适应二维DOA估计研究
发布时间：2018-09-19 19:26
【摘要】：波达方向(DOA)估计是阵列处理中一个常见的任务。传统的多重信号分类(MUSIC)是基于搜索的方法且通常计算成本较高,特别是在联合方位角和俯仰角估计的应用中。在本文中,我们将采用流形分离技术(MST),并提出了一个自适应的二维测向框架,以跟踪任意阵列结构的多个运动目标。首先,我们采用了子空间跟踪技术,每当有新的快拍数据到来时就递归更新一次特征值;此外,通过采用移位矩阵快速单步操作可以并行更新系数矩阵;最后,执行2-D快速傅立叶变换(FFT)来实时计算2-D空间谱。与传统的MUSIC或Root-MUSIC方法相比,基于FFT的方法很大程度上减轻了计算负担,且易于硬件实现。本论文具体的内容安排如下:首先,本文简要概述了自适应二维DOA估计的研究进展,并研究了天线阵列的结构模型,对阵列信号理想情况、存在信号误差及存在相干信源和分布式信源下的矩阵模型进行了分析。其次,我们还介绍了传统的DOA估计算法,其中包括经典的MUSIC与Root-MUSIC算法的实现及优缺点分析对比。并研究了一维情况下流形分离技术在Root-MUSIC方面的改进算法:快速Root-MUSIC与基于IDFT的Root-MUSIC。然后还对二维情况下的两种DOA估计算法进行了对比分析。最后,给出了本论文的核心算法:基于FFT的自适应二维DOA估计算法。该方法通过对系数矩阵的分析,将它分解为了两部分,其一:时间无关项仅需执行一次即可离线获得;其二,时间有关项采用了子空间跟踪技术中的快速近似功率迭代(FAPI)方法来求解权重矩阵,并引入移位矩阵实现了系数矩阵由二步求解转换成更快的单步并行求解的更新过程。接着还将二维空域谱计算转换为了二维FFT操作,从而更快的搜索出对应DOA估计值。最后给出了算法的仿真分析及对比来验证该自适应算法不仅能够减小计算量,而且利于硬件实现。
[Abstract]:DOA (DOA) estimation is a common task in array processing. The traditional multi-signal classification (MUSIC) is a search-based method with high computational cost, especially in the joint azimuth and pitch angle estimation applications. In this paper we use manifold separation technique (MST), and propose an adaptive two-dimensional direction-finding framework to track multiple moving targets with arbitrary array structures. First of all, we use subspace tracking technique to update the eigenvalues recursively whenever new rapid-beat data arrives. In addition, the coefficients matrix can be updated in parallel by using the shift matrix fast one-step operation. The 2-D fast Fourier transform (FFT) is performed to calculate the 2-D spatial spectrum in real time. Compared with the traditional MUSIC or Root-MUSIC methods, the FFT based method greatly reduces the computational burden and is easy to implement in hardware. The main contents of this thesis are as follows: firstly, the research progress of adaptive two-dimensional DOA estimation is briefly summarized, and the structure model of antenna array is studied, which is ideal for array signal. The matrix model with signal error and coherent source and distributed source is analyzed. Secondly, we also introduce the traditional DOA estimation algorithms, including the implementation of the classical MUSIC and Root-MUSIC algorithms and the comparison of their advantages and disadvantages. The improved Root-MUSIC algorithm of manifold separation in one dimension is studied: fast Root-MUSIC and Root-MUSIC. based on IDFT. Then the two DOA estimation algorithms in two-dimensional case are compared and analyzed. Finally, the core algorithm of this thesis is presented: adaptive two-dimensional DOA estimation algorithm based on FFT. By analyzing the coefficient matrix, the method decomposes it into two parts: one is that the time-independent term can be obtained offline only once, the other is that the time-independent term can be obtained offline. The time-dependent term is solved by using the fast approximate power iterative (FAPI) method in subspace tracking technique, and the shift matrix is introduced to realize the updating process of the coefficient matrix from two-step solution to faster one-step parallel solution. Then the two-dimensional spatial domain spectral calculation is converted to two-dimensional FFT operation, so that the corresponding DOA estimation can be searched more quickly. Finally, the simulation analysis and comparison of the algorithm are given to verify that the adaptive algorithm can not only reduce the amount of computation, but also facilitate the hardware implementation.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN911.7
本文编号：2251095
[Abstract]:DOA (DOA) estimation is a common task in array processing. The traditional multi-signal classification (MUSIC) is a search-based method with high computational cost, especially in the joint azimuth and pitch angle estimation applications. In this paper we use manifold separation technique (MST), and propose an adaptive two-dimensional direction-finding framework to track multiple moving targets with arbitrary array structures. First of all, we use subspace tracking technique to update the eigenvalues recursively whenever new rapid-beat data arrives. In addition, the coefficients matrix can be updated in parallel by using the shift matrix fast one-step operation. The 2-D fast Fourier transform (FFT) is performed to calculate the 2-D spatial spectrum in real time. Compared with the traditional MUSIC or Root-MUSIC methods, the FFT based method greatly reduces the computational burden and is easy to implement in hardware. The main contents of this thesis are as follows: firstly, the research progress of adaptive two-dimensional DOA estimation is briefly summarized, and the structure model of antenna array is studied, which is ideal for array signal. The matrix model with signal error and coherent source and distributed source is analyzed. Secondly, we also introduce the traditional DOA estimation algorithms, including the implementation of the classical MUSIC and Root-MUSIC algorithms and the comparison of their advantages and disadvantages. The improved Root-MUSIC algorithm of manifold separation in one dimension is studied: fast Root-MUSIC and Root-MUSIC. based on IDFT. Then the two DOA estimation algorithms in two-dimensional case are compared and analyzed. Finally, the core algorithm of this thesis is presented: adaptive two-dimensional DOA estimation algorithm based on FFT. By analyzing the coefficient matrix, the method decomposes it into two parts: one is that the time-independent term can be obtained offline only once, the other is that the time-independent term can be obtained offline. The time-dependent term is solved by using the fast approximate power iterative (FAPI) method in subspace tracking technique, and the shift matrix is introduced to realize the updating process of the coefficient matrix from two-step solution to faster one-step parallel solution. Then the two-dimensional spatial domain spectral calculation is converted to two-dimensional FFT operation, so that the corresponding DOA estimation can be searched more quickly. Finally, the simulation analysis and comparison of the algorithm are given to verify that the adaptive algorithm can not only reduce the amount of computation, but also facilitate the hardware implementation.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN911.7
【引证文献】
相关期刊论文 前1条
1 汪松年;尹立;闫耀峰;;短波圆环自适应抗干扰研究[J];舰船电子工程;2017年06期
，本文编号：2251095
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2251095.html
