Accurate polynomial chaos expansion for variability analysis using optimal design of experiments

This paper presents a novel generalized polynomial chaos approach for quantifying the uncertainty in high-speed networks arising from random variations in the circuit parameters. The key feature of this work is the development of a non-intrusive linear regression methodology that is able to accurately evaluate the polynomial chaos coefficients using only a sparse set of nodes located within the multidimensional random space. These non-intrusive regression nodes are extracted using an optimized design of experiments (DoEs) approach based on the concepts of exchange algorithms and D-optimality criteria commonly applied in the field of estimation theory.