Some empirical investigations of the autocorrelation function extrapolation
Nezameddin Faghih · Journal of Discrete Mathematical Sciences and Cryptography · 1998
This paper undertakes some empirical investigations of the maximum determinant method of autocorrelation function extrapolation. The digital computer simulations of the first and second order Gaussian processes, with known autocorrelation functions, are used for the empirical investigations; truncated discrete autocorrelation functions are estimated from the data and used to extrapolate the functions and, consequently, estimate the spectral density functions. It provides some practical experience on the extrapolation method and considers some practical factors, emerging in the application of the method to experimentally obtained autocorrelations, revealing interesting features of the method. The reliability test, relating to the truncation point, and the extrapolation errors are considered. The best choice of the truncation point, from where a satisfactory extrapolation may take place, is then deduced. Some exact autocorrelation function simulations are also used for further demonstrations.