A Fast and Accurate Monte Carlo Method for Interconnect Variation

Abstract

For exploring the impact of manufacturing variation on interconnect characteristics, the basic Monte Carlo Method is accurate but computationally very expensive. To overcome the inherent speed limitation we developed an uncomplicated method employing the importance sampling technique. Using confidence intervals our results always take uncertainty into account. The application to a two dimensional interconnect model shows that our method is 23~93 times faster than the basic Monte Carlo method