In this article, an analysis to examine the influence of noise on the global accuracy of the Power Spectral Density (PSD) estimate is made; the expressions of the two components of error caused by noise have been derived, both for arithmetic and geometric mean of smoothed periodograms. It is proved that the performance of non-linear averaging is better than linear one as it produces, approximately, unbiased estimates in presence of noise. Moreover, the effect on accuracy of the PSD estimates, caused by a lack in synchronization with the deterministic component of the signal, are investigated and the exact expression of this error has been derived. The experimental validations seem to definitely confirm these results.
A study on noise-dependent bias error in identification of the parameters of multitone random signals / Attivissimo, Filippo; Savino, Mario; Trotta, Amerigo. - (1995), pp. 149-153. (Intervento presentato al convegno IMEKO TC-4 tenutosi a Praga nel 13-14 Settembre).
A study on noise-dependent bias error in identification of the parameters of multitone random signals
ATTIVISSIMO, Filippo;SAVINO, Mario;TROTTA, Amerigo
1995-01-01
Abstract
In this article, an analysis to examine the influence of noise on the global accuracy of the Power Spectral Density (PSD) estimate is made; the expressions of the two components of error caused by noise have been derived, both for arithmetic and geometric mean of smoothed periodograms. It is proved that the performance of non-linear averaging is better than linear one as it produces, approximately, unbiased estimates in presence of noise. Moreover, the effect on accuracy of the PSD estimates, caused by a lack in synchronization with the deterministic component of the signal, are investigated and the exact expression of this error has been derived. The experimental validations seem to definitely confirm these results.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
