On the Distribution of the Stationary Point of Significance Level for Empirical Distribution Function

We consider empirical distribution functions of nonstationary time-series, depending on set length. The local self- consistent significance level is introduced. The class of time- series, for which the distribution function of significance level is stationary, is considered. For example, the signal-to-interference ratio for random walking subscribers in D2D model of wireless connection belongs to this class of random processes. We introduce also the so-called Chernoff equivalence of the self-consistent significance level and derive the formula of averaging levels for various sets.