A new time-varying autoregressive modeling has been proposed as a tool for time-frequency analysis of nonstationary time series. The method allows a good estimation of both frequency and amplitude of the spectrum and offers a new point of view for the evaluation of the parametric approach when applied to spectral analysis. The good performance is related to the adaptive choice of time-varying coefficients of the autoregressive model. The problems in parametric spectral analysis related to the unsatisfactory amplitude estimate have been successfully overcome by means of a compensation procedure inserted in the designed technique. The evaluation of the proposed approach has been performed on both synthetic and real signals which have been chosen as significant examples of typical nonstationary data. The performance of the method has been compared to classical techniques such as short-time Fourier analysis and autoregressive Burg algorithm. The results show the usefulness of the proposed approach in overcoming many of the difficulties encountered in current spectral estimation techniques by means of parametric approaches. (C) 1999 Elsevier Science B.V. All rights reserved.
A new time-varying autoregressive modeling has been proposed as a tool for time-frequency analysis of nonstationary time series. The method allows a good estimation of both frequency and amplitude of the spectrum and offers a new point of view for the evaluation of the parametric approach when applied to spectral analysis. The good performance is related to the adaptive choice of time-varying coefficients of the autoregressive model. The problems in parametric spectral analysis related to the unsatisfactory amplitude estimate have been successfully overcome by means of a compensation procedure inserted in the designed technique. The evaluation of the proposed approach has been performed on both synthetic and real signals which have been chosen as significant examples of typical nonstationary data. The performance of the method has been compared to classical techniques such as short-time Fourier analysis and autoregressive Burg algorithm. The results show the usefulness of the proposed approach in overcoming many of the difficulties encountered in current spectral estimation techniques by means of parametric approaches. (C) 1999 Elsevier Science B.V. All rights reserved.
Conforto, S., D'Alessio, T. (1999). Optimal estimation of power spectral density by means of a time-varying autoregressive approach. SIGNAL PROCESSING, 72(1), 1-14 [10.1016/S0165-1684(98)00158-3].
Optimal estimation of power spectral density by means of a time-varying autoregressive approach
CONFORTO, SILVIA;D'ALESSIO, Tommaso
1999-01-01
Abstract
A new time-varying autoregressive modeling has been proposed as a tool for time-frequency analysis of nonstationary time series. The method allows a good estimation of both frequency and amplitude of the spectrum and offers a new point of view for the evaluation of the parametric approach when applied to spectral analysis. The good performance is related to the adaptive choice of time-varying coefficients of the autoregressive model. The problems in parametric spectral analysis related to the unsatisfactory amplitude estimate have been successfully overcome by means of a compensation procedure inserted in the designed technique. The evaluation of the proposed approach has been performed on both synthetic and real signals which have been chosen as significant examples of typical nonstationary data. The performance of the method has been compared to classical techniques such as short-time Fourier analysis and autoregressive Burg algorithm. The results show the usefulness of the proposed approach in overcoming many of the difficulties encountered in current spectral estimation techniques by means of parametric approaches. (C) 1999 Elsevier Science B.V. All rights reserved.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
