1992 年 2 巻 2 号 p. 137-144
The approximate Cramer-Rao lower bound (CRLB) for unbiased parameter estimators of discrete fractional Gaussian noise is derived on the basis of the asymptotic diagonalization of covariance matrix. Since the approximate CRLB can be computed using the FFT method, it requires an extremely cheaper computational cost than the exact CRLB which is efficiently calculated using the Levinson algorithm. It is found from numerical experiments that the approximate CRLB agrees well with the exact CRLB for N (number of data points)≳10^2.