De-Warping Algorithms for Oscillatory Almost-Cyclostationary Processes

Recently, the class of the oscillatory almost-cyclostationary (OACS) processes has been introduced and characterized in the wide sense. These processes have a Priestley spectral representation where the underlying spectral process is almost cyclostationary. In this paper, OACS processes are characterized in the strict sense. For the special case of the amplitude-modulated time-warped almost-cyclostationary processes, a recently proposed estimation algorithm of the time-warping function, which is based on the structure of the autocor-relation function, is generalized under less restrictive assumptions to exploit the structure of first-order statistical functions. Then, de-warping procedures that exploit these estimates are carried out. Numerical experiments made on simulated data show the effectiveness of the proposed methods. Moreover, numerical results on a real electrocardiogram signal validate the effectiveness of the recently proposed OACS model for this signal.