一种不需要特征值分解的ESPRIT方法

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。

An Improved ESPRIT Algorithm without Eigenvalues Decomposition

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 filter is brought up.The suitable con- figuration of multi-stage Wiener filter for ESPRIT is proposed.The new method has no need of the eigenvalues decomposition of the auto- correlation 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.