Title :
Robust M-periodogram
Author :
Katkovnik, Vladimir
Author_Institution :
Dept. of Stat., South Africa Univ., Pretoria
fDate :
11/1/1998 12:00:00 AM
Abstract :
A maximum likelihood-type M-periodogram is developed for observations contaminated by impulse random errors having unknown heavy-tailed error distributions. The periodogram is defined as being a squared amplitude of a harmonic signal fitting the observations with, a nonquadratic residual loss function given by the Huber´s (1981) minimax robust statistics
Keywords :
error statistics; maximum likelihood estimation; minimax techniques; noise; signal sampling; deterministic signal; harmonic signal fitting; heavy-tailed error distributions; impulse random errors; maximum likelihood-type M-periodogram; minimax robust statistics; noisy signal samples; nonquadratic residual loss function; observations; robust M-periodogram; squared amplitude periodgram; Africa; Degradation; Gaussian distribution; Maximum likelihood estimation; Minimax techniques; Power harmonic filters; Probability distribution; Robustness; Spectral analysis; Statistical distributions;
Journal_Title :
Signal Processing, IEEE Transactions on
