An Efficient Three-Stage Yield Optimization Technique for Analog Circuits Using Evolutionary Algorithms

In addition to the improved calculation of the parameter values, a high yield estimation is necessary for designing analog integrated circuits. Although Monte-Carlo (MC) simulation is popular and precise for yield estimation; however, its efficiency is not high enough and it requires too many costly transistor-level simulations. Therefore, some accelerated methods are needed for MC simulations. This paper presents a novel approach for improving automated analog yield optimization using a three-stage strategy. Firstly, critical solutions are recognized using Critical Analysis (CA) and Multi-objective Optimal Computing Budget Allocation (MOCBA). Then they are separated from non-critical answers. It's so helpful to avoid repeating the Monte Carlo (MC) simulations of non-critical solutions. Due to the existence of several objective functions (typically more than one) in the yield optimization problem, by using the Multi-Objective Optimization (MOO) in the second stage, more precise answers can be found. Finally, MC simulations are performed to explore the proposed algorithm performance. Simulation results show that our approach locates higher quality in terms of yield rate within less run time and without affecting the accuracy.