ABSTRACT
When the noise is nonwhite (or colored), a proper assessment of the noise spec-
trum is beneficial to the detection and estimation of the sinusoidal signals, es-
pecially for small sample sizes. In some applications, noise-only samples are
independently available so that the noise spectrum can be estimated in the ab-
sence of sinusoids. In other applications, the noise spectrum has to be estimated
from a time series that possibly contains sinusoidal components with unknown
parameters. In this chapter, we discuss some techniques of spectral estimation
under both scenarios, but with a special emphasis on the latter case. We pay
attention to both real-and complex-valued time series.