Frequency correlation analysis for periodic echoes

This paper presents a multiplicative model for a class of nonstationary signals, referred to as cyclostationary, along with a frequency correlation (FC) analysis particularly suited for such signals. Signals of this class can be found in the ultrasonic imaging of certain materials and signals used in communication systems. This paper applies an FC analysis to detect the mean scatterer spacing of a sequence of quasi-periodic scatterers with random scatterer strengths. Mathematical derivations and numerical experiments are presented to demonstrate the utility and feasibility of FC methods for analyzing signals created by the multiplication of random processes. The results show that FC methods can outperform simple power spectral density and the cepstral methods in analyzing cyclostationary signals.