ABSTRACT
For a time series that comprises sinusoids in white noise, the autocovariance
function (ACF) has a very special structure: it is a weighted sum of sinusoids,
whose frequencies are the same as the frequencies of the sinusoidal signal, plus
a Kronecker delta sequence. In this chapter, we discuss several methods of fre-
quency estimation that utilize the special structure through the application of
a powerful technique of matrix analysis — the eigenvalue decomposition (EVD).
The key idea is to construct suitable matrices based on the ACF such that the
signal frequencies can be obtained from their eigenvalues and eigenvectors.