POWER SPECTRUM ESTIMATION FROM NOISY AND LIMITED AUTOCORRELATION VALUES - A MAXIMUM-ENTROPY APPROACH

The problem of estimating the power spectrum from uncertain and limite d autocorrelation values is considered in this paper. It is proposed t hat in order to reduce noise effects in the estimate, oversampling of the available time domain of the data be carried out. The oversampling problem is discussed within a new framework, with a maximum entropy a lgorithm appropriate to the case when a smoothness property of the pow er spectrum is required. (C) 1997 Elsevier Science B.V.