Monte Carlo based statistical timing analysis using an efficient sampling method

Abstract

Statistical static timing analysis (SSTA) is one of the most commonly used methods for analyzing timing behavior under process variations (PVs). Monte Carlo (MC) simulation is a straightforward and effective method for SSTA. In this paper, a new sampling method is proposed to improve an MC-based approach by reducing the number of simulation samples. This method utilizes stratification with an importance analysis for the selection of samples of PVs. Extensive simulations are performed to show that the proposed method achieves a better precision on ISCAS 89 benchmarks compared to other MC simulation approaches.