序列二次逼近下的集成电路统计最优化方法

基于对电路特性函数的序列正交二次逼近，本文给出了考虑工艺参数随机起伏引起电路特性随机变化的统计最优化方法。该方法是根据集成电路设计和制造的特殊性提出的，它与传统的统计最优化模型和方法有显著区别，适合于电路设计参数和工艺扰动参数不在同一空间的最优化问题。本文给出了这一模型的理论框架及其证明，并结合ＣＭＯＳ电路的最优设计证明了该方法的可行性和实用性，得到了满意的结果。

IC Statistical Optimization Technology with Sequential Quadratic Approximation

Based on the sequential quadratic approximation medel of circuit characteristics functions,a new statistical optimization technology is given in which the random fluctuation of circuit performance is considered.Because of the specialities during IC design and manufacture,the method solves parameter optimal problem,of which design space and fluctuation space are different.In this paper,the theoretic framework of this model is obtained.The feasibility and utility of the method are demonstrated satisfactorily by an example of CMOS design at last.