Title :
Spectral analysis using prediction methods
Author_Institution :
FGAN-Forschungsinstitut für Hochfrequenzphysik, Wachtberg-Werthhoven
Abstract :
The presented method is an extension of the well known autoregressive or Maximum-Entropy-Method. Here the power spectrum of the given process is fitted to a rational function of degree d. A motivation is given why it is better to use rational function instead of functions of the Form 1/ | Polynom (λ)|2. Some examples illustrate the use of the method. No information is given about the statistics of the estimate.
Keywords :
Extrapolation; Length measurement; Nonlinear filters; Prediction methods; Random variables; Signal processing; Spectral analysis; Statistics; Time measurement; Transfer functions;
Conference_Titel :
Acoustics, Speech, and Signal Processing, IEEE International Conference on ICASSP '79.
DOI :
10.1109/ICASSP.1979.1170767