A new class of complex time-frequency distributions for estimating phase synchrony between signals

Determining the time-varying phase spectrum of non-stationary signals is important for the quantification of the dynamics between signals and has been widely used in detecting synchrony between chaotic oscillators. Current work in quantifying time and frequency dependent phase information relies on either the Hilbert transform or the complex Morlet wavelet transform. Although these methods are effective at extracting the time-varying phase information, they have some drawbacks such as the assumption that the signals are narrowband for the Hilbert transform and the non-uniform timefrequency resolution inherent to the wavelet analysis. In this paper, we propose using a general class of complex distributions belonging to Cohen's class for defining time-varying phase spectrum and phase synchrony. This new class of distributions is defined using the spectrogram decomposition of time-frequency distributions and is shown to have improved performance in detecting phase synchrony compared to existing methods using simulated signals.