Componentwise coherence function of correlated periodically nonstationary random processes

Abstract

In this paper we consider a novel componentwise coherence function, which is determined by the cross-spectral densities of stationary components of periodically nonstationary random processes i.e. stationary correlated random processes, which modulate their carrier harmonics. The properties of the introduced coherence function are specified for the amplitude- and phase-modulated signals. Its graphical frequency dependencies have been obtained for the specified signal parameters. The advantages of componentwise coherence function in comparison with previously suggested integral coherence function are shown. We present the method for the selection of stationary modulating components, which is based on the frequency shift and low-pass filtration. The properties of selected components for the cases of amplitude- and phase-modulated processes are analyzed.