Measure of cyclostationarity for Gaussian processes based on the likelihood ratio test

The problem addressed in this paper is the detection of cyclostationarity, and the measurement of the trend of a process to have this property. This problem is of great importance, because in applications algorithms using the property of cyclostationarity assume the periodicities of the statistics to be known. Thus the periodicities need to be detected/estimated, and furthermore, a measure must be given in order to qualify the trend of a process to have a given periodicity. This measure will give information about the opportuneness of using a cyclostationary modelization instead of a stationary one.