Comparison of nonparametric and parametric time-varying methods for quantifying phase synchrony

Quantifying the pairwise relationships between two signals is an important problem in many fields of science and engineering. Phase synchrony has been suggested as a powerful tool for determining bivariate signal relationships. Recently, a nonparametric phase estimation method based on the Rihaczek distribution belonging to Cohen´s class of time-frequency distributions (TFDs) has been proposed for the estimation of time-varying phase. Alternatively, parametric time-frequency methods, which model the phase as a polynomial, can be used. In this paper, the nonparametric phase estimator based on the Rihaczek distribution is compared with the parametric estimator based on the high-order ambiguity function for their accuracy in estimating phase synchrony.