Efficient Global Optimization of Analog Circuits Using Predictive Response Surface Models on Discretized Design Space

摘要

Analog circuit optimization is an important-yet-challenging problem, because accurately evaluating circuit performance by transistor-level simulation within the optimization loop is extremely expensive. This article addresses the grand challenge by adopting two novel techniques: first, discretizing the continuous design space with respect to the variability-induced correlation, making the total number of design candidates finite; second, performing circuit simulation only for the top design candidates determined by response surface models. The efficacy of the proposed method is demonstrated by several practical circuit examples.