稀疏欠采样阵列信号处理方法研究

发布时间：2018-01-04 11:42

本文关键词：稀疏欠采样阵列信号处理方法研究　出处：《电子科技大学》2014年博士论文　论文类型：学位论文

【摘要】：压缩感知理论是一种利用信号的稀疏特性从欠采样测量值中恢复信号的技术，该理论不受传统Nyquist采样定理约束，因而具有广泛的学术研究和工业应用价值。目前，压缩感知理论重点研究了特定字典情况下的稀疏信号重建算法模型、重建条件等，而针对压缩感知理论在阵列信号处理中的具体应用仍有许多待深入研究的内容。过水平采样是另一种降低采样率的方法，其也被称为事件驱动采样，，即由信号本身决定何时采样。过水平采样可以在获取信息的同时有效地减少采样样本数，实现高效节能地获取、探测与估计信息的目的。但目前过水平采样理论、信号重建算法及其在信号处理中的具体应用还不成熟。极化阵列既可以获得信号的空间信息，又可以获得信号的极化信息，完备的电磁信息为阵列性能的提高奠定了物理基础，使得在稀疏极化域条件下进一步改善阵列信号处理性能成为可能。但现有针对极化阵列的波达方向（Direction-of-Arrival, DOA）估计算法因待估参数多而存在计算量大的问题。针对以上情况，本文重点研究了基于压缩感知理论的应用算法、压缩感知理论和过水平采样理论的联合优化算法，以及针对极化阵列的信号处理相关算法。本论文的主要工作如下： 1.针对非相关信号和相干信号同时存在的情况下，提出一种基于信号子空间方法和子空间块稀疏重建理论的DOA估计方法。所提算法可以有效地估计信号的DOA、不受阵列几何结构的限制，且具有超载能力。进一步研究了色噪声环境下的非高斯信号DOA估计问题，提出了基于四阶累积量(Fourth-order Cumulants, FOC)的DOA估计算法。另外，在利用阵列方向分集的情况下，提出了一种基于双约束弹性树搜索正交匹配追踪的DOA估计算法。 2.提出一种基于稀疏重建理论和过水平采样理论的联合优化DOA估计算法。首先针对模拟信号进行过水平采样，并研究了该欠采样方案可以利用简单的1-比特模数转换器硬件电路实现。同时基于欠采样测量数据，将DOA估计问题描述为压缩感知稀疏重建框架下的一个代价函数。该算法可以降低阵列信号处理系统的采样速率和硬件成本。 3.提出一种基于稀疏重建理论的波束综合算法，首先将压缩感知的稀疏重建理论和凸优化引入到线性阵列的波束综合问题。进一步研究基于迭代重加权1范数最小化和凸优化的阵列波束图综合算法，并将这一方法拓展到由阵列方向性所构建的二维阵列的波束综合问题，最后研究了关于极化阵列的波束综合问题及基于稀疏表示的圆阵波束形成问题。 4.提出一种针对稀疏极化域阵列的低复杂度DOA估计算法。首先引入了部分校正分布式电磁矢量传感器阵列的测量模型，并推导出由关于四维参数退化到关于二维参数的谱搜索代价函数。同时将该低复杂度DOA估计方法拓展到关于未校正极化阵列的DOA估计问题。最后介绍了一种类似ESPRIT的低复杂度DOA估计算法，并给出了理论分析。
[Abstract]:Compressed sensing theory is a kind of sampling measurement technology to restore the signal from the less use of signal sparse characteristics, this theory is not affected by the traditional Nyquist sampling theorem constraint, thus it has a wide range of academic research and industrial application value. At present, the research model of sparse signal reconstruction algorithm specific dictionary under the theory of compressed sensing, reconstruction conditions so, the specific application of compressed sensing theory in array signal processing is still a lot to be studied.
Horizontal sampling is another method to reduce the sampling rate, which is also known as event driven by the sampling signal itself determines when sampling. Over the level of access to information at the same time sampling can effectively reduce the sample number, efficiency gains, detection and estimation of information. But the current level of sampling the theory of signal reconstruction algorithm and its application in signal processing is not mature.
Polarized array can get the spatial information signal, and can obtain the polarization information signal, complete information to improve the electromagnetic performance of the array provided the physical basis, the conditions to further improve the polarization domain sparse array signal processing becomes possible. But the existing performance for polarized array DOA (Direction-of-Arrival, DOA) estimation algorithm for the estimated parameters and the problems of large amount of calculation.
In view of the above situation, this paper focuses on the application algorithm based on compressed sensing theory, the joint optimization algorithm of compressive sensing theory and over level sampling theory, and the related algorithm of signal processing for polarization array.
1. for uncorrelated signals and coherent signals exist at the same time, a method is proposed based on the estimation of signal subspace method and subspace block sparse reconstruction theory DOA. The proposed algorithm can effectively estimate the signal from DOA, array geometry constraints, and has overload ability. Further study of the non Gauss signal DOA colored noise estimation is proposed based on four order cumulant (Fourth-order, Cumulants, FOC) DOA estimation algorithm. In addition, in the case of using array diversity, we propose a double elastic constraint tree search based on orthogonal matching pursuit DOA estimation algorithm.
2. proposed a joint optimization of DOA based on sparse reconstruction theory and theory of horizontal sampling estimation algorithm. Firstly the analog signal level sampling, and studied the undersampling scheme can be implemented by hardware circuit of 1- bit analog-to-digital converter is simple. At the same time sampling based on measured data, the DOA estimation problem is described as a price compressed sensing sparse reconstruction under the framework of function. This algorithm can reduce the sampling rate and the hardware cost of array signal processing system.
3. propose a pattern synthesis algorithm based on sparse reconstruction theory, the compressed sensing sparse reconstruction theory and convex optimization problem into the pattern synthesis of linear array. Further research on the iterative reweighted 1 minimization and convex optimization algorithm based on array beam pattern synthesis, and this method is extended to two-dimensional array pattern synthesis problem constructed by the array direction of the problem, formed at the end of the beam synthesis problems about polarization array and circular array beamforming based on sparse representation.
4. we propose a low complexity DOA for sparse array polarization domain estimation algorithm. Firstly, partial correction of distributed electromagnetic vector sensor array measurement model, and deduced by a four-dimensional parameter degradation to spectral search cost function on the two-dimensional parameters. At the same time the low complexity DOA estimation method is extended to a polarization correction array DOA estimation problem. Finally introduces the low complexity DOA a similar ESPRIT estimation algorithm, and the theoretical analysis is given.

【学位授予单位】：电子科技大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN911.7

【参考文献】