An Improved Multi-Objective Optimization Framework for Soft-Error Immune Circuits

Abstract

Soft error is one of the main circuit reliability issues. Mitigating soft error inevitably requires sacrificing area and power, therefore, it is necessary to balance area, power, and soft error. In this paper, some improvements have been made to the multi-objective optimization framework based on Back Propagation (BP) neural network and Non-dominated Sorting Genetic Algorithm-II (NSGA-II). A data set selection and dimensionality reduction scheme is proposed to ensure that the framework is suitable for circuit designs of different scales. The experimental results show that the average soft error rate (SER) of the five circuits is reduced by 47.6%, the area is increased by 12.1%, and the power is increased by 31.5%.