Introduction
The modelling concerns either the system generating the observation whose PSD we want to estimate: parametric methods, or the signal itself: maximum likelihood method [15].
Parametric spectral estimation methods use a model to obtain an estimate of the spectrum. Based on a priori knowledge of the process, these models fall into three main categories:
Autoregressive (AR) models
Adjusted mean (MA) models
autoregressive average-adjusted (ARMA) models.
We break down the parametric approach into three steps:
Choose a model that describes the process appropriately.
Estimate model parameters based on available data.
Estimate the spectrum based on the model parameters.