Phase coupling detection using a finite number of realizations and non-independent phases

Phase coupling detection and quantification by using the bispectrum is theoretically attractive and easy to understand. However some problems appear in practice, especially the bispectrum estimation on a finite (and small) trials number implies false phase coupling detection and bad quantification. Moreover the phase independence which is assumed, for each realization, is not always verified. So, in this paper we discuss an alternative method of phase coupling detection taking into account these two limitations