一种不需要特征值分解的ESPRIT方法

An Improved ESPRIT Algorithm without Eigenvalues Decomposition

多级维纳滤波是一种降维算法,本文提出了一种将多级维纳滤波引入到ESPRIT算法的预处理模型,提出了一种构造预处理滤波器的方法,提出了适合于ESPRIT方法的多级维纳滤波的结构。本文提出的方法避免了采样数据二阶统计的特征值分解,降低了运算量。本文方法具有一般性,适合于基于子空间方法的一大类DOA估计。仿真结果证明了方法的有效性。

Multi-stage Wiener filter is a reduced-rand algorithm. In this Paper, the pretreatment model is presented, thus the multi-stage Wiener filter can be used in ESPRIT algorithm. A method to construct the pretreatment fiher is brought up. The suitable configuration of multi-stage Wiener filter for ESPRIT is proposed. The new method has no need of the eigenvalues decomposition of the autocorrelation of the sample data,thus can have less computational cost. The new method can apply to a lot of subspace methods. Simulation results demonstrate the effectiveness of the method.