OCBA in the yield optimization of analog integrated circuits by evolutionary algorithms

An strategy based on the optimal computing budget allocation (OCBA) approach is presented to reduce the simulation cost in the yield optimization of analog integrated circuits (ICs), when it is performed on classical Monte Carlo simulations. OCBA is applied to distribute the large portion of the budget simulations among critical IC sizing optimization cases, while limiting the simulations for the non-critical cases. The optimal sizing is performed by applying three evolutionary algorithms, namely: non-dominated sorting genetic algorithm (NSGA-II), multi-objective evolutionary algorithm with decomposition (MOEA/D), and particle swarm optimization (PSO). They are executed within the OCBA-based strategy to enhance the yield of an operational transconductance amplifier using IC technology of 90 nm.