基于稀疏贝叶斯学习的DOA估计算法

发布时间：2018-10-16 12:54

【摘要】：波达方向角(DOA)估计技术在许多实际应用中扮演着十分重要的角色,因此一直是一个备受关注的研究领域。传统的DOA估计算法需要大量的采样数据、在较高的信噪比条件下才能精确估计,而且处理相干信号效果不理想,在一些新兴领域的应用中受到了很大的限制。因此,如何利用少量的测量数据来实现高分辨率的DOA估计算法成为一个新兴的研究方向。稀疏表示理论的不断发展为DOA估计提供了新的发展方向,它指出对于稀疏或可压缩信号,当满足一定条件时,可以用远低于Nyquist采样定理所需要的采样数据精确地恢复出原始信号。主流的稀疏重构算法有三类,分别是贪婪算法,凸松弛类算法和贝叶斯学习算法。其中贝叶斯学习/推断综合了决策者的先验认知、样本信息中未知参数的分布信息,从统计优化的角度达到稀疏重构的目的,容易理解而且表现出许多优点,现已成为一个研究热点。本文重点研究了基于稀疏贝叶斯学习的DOA估计算法,主要工作和成果列举如下:1.详细介绍了三类稀疏重构算法的基本原理并分析了各自的优缺点,对经典的基于稀疏表示的DOA估计算法做了简单介绍,其中包括L1-SVD算法、L1-SRACV算法和降维的L1-SRACV算法。2.阐述了贝叶斯估计理论、贝叶斯学习算法的框架,分析了贝叶斯学习算法的优势,最后,提出了基于多测量矢量稀疏贝叶斯学习(MSBL)的DOA估计算法,通过仿真实验证明了该算法在低快拍情况下有很好的估计性能,并且可以应用于相干信号的处理。3.研究学习了分布源DOA估计的信号模型和分块稀疏贝叶斯学习(BSBL)算法,将BSBL应用于分布式信号源的DOA估计中,通过仿真测试分析了该算法的估计性能。
[Abstract]:Direction-of-arrival (DOA) estimation plays an important role in many practical applications, so it is always a concerned field. The traditional DOA estimation algorithm needs a large number of sampled data to estimate accurately under the condition of high signal-to-noise ratio (SNR). Moreover, the effect of processing coherent signal is not ideal, so it is limited in some new applications. Therefore, how to use a small amount of measurement data to achieve high resolution DOA estimation algorithm has become a new research direction. The continuous development of sparse representation theory provides a new direction for DOA estimation. It points out that for sparse or compressible signals, when certain conditions are satisfied, The original signal can be accurately recovered from the sampling data that is far less than that required by the Nyquist sampling theorem. There are three main sparse reconstruction algorithms: greedy algorithm, convex relaxation algorithm and Bayesian learning algorithm. Bayesian learning / inference integrates the prior cognition of decision makers, the distribution of unknown parameters in sample information, and achieves the purpose of sparse reconstruction from the angle of statistical optimization. It is easy to understand and shows many advantages. It has become a research hotspot. This paper focuses on the DOA estimation algorithm based on sparse Bayesian learning. The main work and results are listed as follows: 1. The basic principle of three kinds of sparse reconstruction algorithms is introduced in detail, and their advantages and disadvantages are analyzed. The classical DOA estimation algorithm based on sparse representation is briefly introduced, including L1-SVD algorithm, L1-SRACV algorithm and dimensionally reduced L1-SRACV algorithm. 2. The Bayesian estimation theory and the framework of Bayesian learning algorithm are expounded. The advantages of Bayesian learning algorithm are analyzed. Finally, a DOA estimation algorithm based on sparse Bayesian learning (MSBL) is proposed. The simulation results show that the proposed algorithm has good estimation performance in the case of low shot and can be applied to coherent signal processing. The signal model of distributed source DOA estimation and block sparse Bayesian learning (BSBL) algorithm are studied. BSBL is applied to the DOA estimation of distributed signal source. The performance of the algorithm is analyzed by simulation test.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN911.7

【相似文献】