空间谱估计经典算法性能比较

The Performance Comparison of Spatial Spectrum Estimation Classic Algorithm

空间谱估计是阵列信号处理的一个重要研究方向。空间谱估计理论与技术已日趋成熟,近几十年的经典谱估计技术包括:常规波束形成(CBF)、Capon谱估计、多重信号分类(MUSIC)、旋转不变子空间算法(ESPRIT)、最大似然(ML)、子空间拟合(SF),及这些算法的扩展和变形。上述算法在各个分散的文章中均有具体深入的理论分析和研究,亦有类似的2种或3种算法的性能比较,但是针对这些所有算法的性能比较,就笔者所知尚无公开报道,而使工程实现时对算法的选择没有依据。本文对这些经典算法做了简介,列出各个算法的优缺点,并对性能进行仿真比较,能直观地得到各个算法的性能对比,给工程实现算法选择提供理论依据。

Spatial spectrum estimation is an important research direction in array signal processing. Spatial spectrum estimation theory and the technology has matured, in recent decades classical spectral estimation techniques include： Conventional beam forming （CBF）, Capon spectral estimation, multiple signal classification （MUSIC）, and rotation invariant sub-space algorithm （ESPRIT）, maximum likelihood （ML） and subspace fitting （SF） and the extension of these algorithms and deformation. The above algorithms have a specific in-depth theoretical analysis and research in various scattered articles, but for all of these algorithms performance comparison as I know there is no publicly reported, and make the choice of algorithm not have the basis while the project implementation. The paper provide a classical algorithm introduction, lists the advantages and disadvantages of each algorithm, and the simulation to the properties which can provide the performance of each algorithm contrast, provides the theory basis of choice for the project implementation algorithm.