A Multi-objective Optimization Framework to Design Soft-Error-Immune Circuit

Abstract

It is widely accepted that Triple Module Redundancy (TMR) hardened method is one of the most effective approach to protect circuit from soft error. However, while the mitigation of Soft Error Rate (SER) entails the inevitable sacrifice of area and power, little attention has been drawn to an optimization method to explore a trade-off between area, power and SER. Thus, with the help of neural network algorithm and NSGA-II algorithm, this study investigates a multi-objective optimization model as a guidance on Soft-Error-Immune circuit design.