Time-frequency statistical characteristics of cyclostationary signals

Time-frequency statistical characteristics of cyclostationary signals are studied. The characteristics considered are real and complex instantaneous autospectra. Unfortunately, as both these characteristics can take on negative values, they are not spectral densities. Moreover, the second characteristics is a complex function of time and frequency. The first characteristic, on the other hand, can be deteriorated by the occurrence of artifacts. The origin of these artifacts is shown to be the difference between the support area, on which the autospectrum is unambiguously defined, and the area, which represents the basic period of the autospectrum. However, both characteristics can be computed with very good time and frequency resolution