基于改进的Capon滤波器组谱估计新方法

Spectral estimation methods based on improved Capon filterbank

为了解决Capon滤波器组谱估计方法在稳定分布噪声下性能出现退化的问题,提出了一种改进的Capon滤波器组谱估计方法,利用分数低阶协方差(FLOC)对传统的Capon滤波器组进行重构。由于分数低阶协方差的作用范围是0<≤2整个分数低阶空间,因此在稳定分布噪声下其谱估计性能比基于二阶统计量方法的谱估计性能更好。计算机仿真结果显示了这种方法的优越性能,并且在高斯噪声环境下,其性能与二阶统计量下的方法类似。

In order to solve the degenerative performance of Capon filterbank spectral estimation under alpha-stable distribution noise, a new spectral estimation method based on improved Capon filtrbank is presented, the traditional Capon filterbank is reconstructed with fractional lower order covariance（FLOC）.Because the fractional lower order covariance is used in 0 ≤2, the whole space of fractional lower order statistics, so the performance of spectral estimation better than the methods based on second order statistics.The simulation shows the new methods based on FLOC outperforms the method based on second order statistics in the alpha-stable noise while performs similar to their respective methods in the Gaussian noise.