An Overview of Classical and High Resolution Spectral Estimation

Current methods of spectral estimation can be broadly categorized under three main headings. One is classical power spectral density estimation which incorporates estimation of the autocorrelation function through lagged products and periodogram analysis and its variations. The second is power spectral density estimation based on modelling. This incorporates maximum entropy analysis, data extension using linear prediction and spectral estimation using ARMA models. The third is power spectral density estimation using adaptive windows which incorporates the method commonly referred to as the maximum likelihood (MLM) method. (Author)