The controlled stratification method to estimate extreme quantiles in the field of EMC modelling

This paper deals with EMC modelling analysis with regard to uncertain input parameters. We introduce a variance reduction technique called controlled stratification to estimate an extreme quantile of an interest response of a complex model. This technique relies on using a simple model to identify relevant input realizations. Then, these realizations allow evaluating the complex model for the quantile estimation. In this communication, this approach is applied to estimate an extreme quantile of a crosstalk problem, thus reducing the computational cost compared to a direct empirical estimation.