交替分离算法的Cramer-Rao界

Alternating Separation Algorithm and Its Cramer-Rao Bound

该文推导了交替分离算法的Cramer-Rao界。交替分离算法的Cramer-Rao界涉及到矩阵的“减”逆,而矩阵“减”逆具有比通常Moore-Penrose广义逆更为宽松的定义条件,在理论上,一个确定矩阵有无数个“减”逆。为了建立分离算法的Cramer-Rao界,该文求出了一个确定矩阵的一个特定“减”逆矩阵。根据任一确定性定理,得到分离数据的密度分布函数,从而获得交替分离算法的Cramer-Rao界。交替分离算法的Cramer-Rao界将多信号对信号参数估计的影响能更直观反映出来。通过对交替分离算法的Cramer-Rao界的讨论,该文还给出了有关矩阵分离算子一些重要的性质。

This paper provides the Cramer-Rao lower bound of alternating separation algorithm. In the Cramer-Rao bound, the ＂-＂ inversion of a definite matrix must be established, which extends the theory of Cramer-Rao bound. In additional, this Cramer-Rao bound displays the multiple signal effect in the parameter estimated process. By discussion the Cramer-Rao lower bound of Alternating Separation algorithm, some properties of the signal separation operator matrix are given.