Mean Performance Optimization of Analog Integrated Circuits via Expectation-Based Stochastic Programming

摘要

In this paper we propose to formulate the mean performance optimization of analog ICs as an expectation-based stochastic program, which can be further approximated as a more tractable optimization problem that interior-point methods can efficiently solve. Using such stochastic optimization approach, we can obtain statistical analog circuit designs with optimized mean performance. A ring oscillator design is given to demonstrate the method.