Single event effects evaluation on convolution neural network in Xilinx 28 nm system on chip

赵旭 ¹杜雪成 ¹熊旭 ¹马超 ¹杨卫涛 ²郑波 ¹周超¹

扫码查看

作者信息

1. School of Nuclear Science and Technology,University of South China,Hengyang 421001,China
2. School of Microelectronics,Xidian University,Xi'an 710071,China
折叠

Abstract

Convolutional neural networks(CNNs)exhibit excellent performance in the areas of image recognition and object detection,which can enhance the intelligence level of spacecraft.However,in aerospace,energetic particles,such as heavy ions,protons,and alpha particles,can induce single event effects(SEEs)that lead CNNs to malfunction and can significantly impact the reliability of a CNN system.In this paper,the MNIST CNN system was constructed based on a 28 nm system-on-chip(SoC),and then an alpha particle irradiation experiment and fault injection were applied to evaluate the SEE of the CNN system.Various types of soft errors in the CNN system have been detected,and the SEE cross sections have been calculated.Furthermore,the mechanisms behind some soft errors have been explained.This research will provide technical support for the design of radiation-resistant artificial intelligence chips.

Key words

single event effects/convolutional neural networks/alpha particle/system on chip/fault injection

引用本文复制引用

基金项目

National Natural Science Foundation of China(12305303)

Natural Science Foundation of Hunan Province of China(2023JJ40520)

Natural Science Foundation of Hunan Province of China(2021JJ40444)

Natural Science Foundation of Hunan Province of China(2019JJ30019)

Research Foundation of Education Bureau of Hunan Province of China(20A430)

Science and Technology Innovation Program of Hunan Province(2020RC3054)

Natural Science Basic Research Plan in the Shaanxi Province of China(2023-JC-QN-0015)

Doctoral Research Fund of University of South China()

出版年

2024

中国物理B(英文版)

中国物理学会和中国科学院物理研究所

中国物理B(英文版)

CSTPCDEI

影响因子：0.995

ISSN：1674-1056

参考文献量25

段落导航