Stochastic Modeling-Based Variability Analysis of On-Chip Interconnects

Abstract

In this paper, a novel stochastic modeling strategy is constructed that allows assessment of the parameter variability effects induced by the manufacturing process of on-chip interconnects. The strategy adopts a three-step approach. First, a very accurate electromagnetic modeling technique yields the per unit length (p.u.l.) transmission line parameters of the on-chip interconnect structures. Second, parameterized macromodels of these p.u.l. parameters are constructed. Third, a stochastic Galerkin method is implemented to solve the pertinent stochastic telegrapher's equations. The new methodology is illustrated with meaningful design examples, demonstrating its accuracy and efficiency. Improvements and advantages with respect to the state-of-the-art are clearly highlighted.