显示器电磁信息泄漏的机器学习检测方法研究

关天敏; 韩振中; 茅剑

引用本文：

关天敏,韩振中,茅剑.显示器电磁信息泄漏的机器学习检测方法研究[J].信息安全学报,2021,6(2):101-109 [点击复制]
GUAN Tianmin,HAN Zhenzhong,MAO Jian.Research on the detection method of electromagnetic information leakage from display by machine learning[J].Journal of Cyber Security,2021,6(2):101-109 [点击复制]

本文已被：浏览 4509次下载 3407次	码上扫一扫！
显示器电磁信息泄漏的机器学习检测方法研究
关天敏^1,2, 韩振中³, 茅剑^4,2
0 字体:加大+\|默认\|缩小-
(1.集美大学信息工程学院厦门中国 361021;2.厦门市涉密信息技术重点实验室厦门中国 361021;3.国防科技大学电子对抗学院合肥中国 230000;4.集美大学计算机工程学院厦门中国 361021)

摘要:

根据电磁学原理，在操作电子信息设备的过程中会产生无意的电磁辐射。电磁辐射会引发信息泄漏，给信息安全造成严重威胁。面向计算机显示器的电磁信息安全问题，提出基于机器学习的电磁信息泄漏检测方法。针对电磁泄漏信号的特点，设计了MGCNN卷积神经网络。利用其独特的卷积和池化处理能力，提取显示器电磁频谱信号中图像信息的多层次特征，克服了传统检测方法需要事前明确电磁信息特征和缺乏自适应能力的缺陷，从而有效地解决电磁信号中的信息泄漏检测问题。通过实测对比，证明了MGCNN对于显示器的电磁信息泄漏检测的有效性。

关键词: 电磁信息泄漏电磁辐射信息安全机器学习卷积神经网络计算机显示器 TEMPEST

DOI：10.19363/J.cnki.cn10-1380/tn.2021.03.07

投稿时间：2020-03-24修订日期：2020-07-08

基金项目:本课题得到福建省自然科学基金资助项目（No.2017J01762）；福建省科技厅重点项目（No.2018H0025）；厦门市科技局资助项目（No.3502Z20183037）资助。

Research on the detection method of electromagnetic information leakage from display by machine learning

GUAN Tianmin^1,2, HAN Zhenzhong³, MAO Jian^4,2

(1.School of Information Engineering, Jimei University, Xiamen 361021, China;2.Xiamen Key Laboratory of secret information technology, Xiamen 361021, China;3.College of Electronic Engineering, National University of Defense Technology, Hefei 230000, China;4.Computer Engineering College, Jimei University, Xiamen 361021, China)

Abstract:

According to the principles of electromagnetics, unintentional electromagnetic radiation is generated during the operation of electronic information equipment. Electromagnetic radiation can cause information leakage and pose a serious threat to information security. Faced with the problem of electromagnetic information security of computer monitors, a method of electromagnetic information leakage detection based on machine learning is proposed. According to the characteristics of electromagnetic leakage signal, MGCNN convolutional neural network is designed. Using its unique convolution and pooling processing capabilities, MGCNN extracts multi-level features of image information in the electromagnetic spectrum signal of the display. It overcomes the defects of traditional detection methods that need to make clear the characteristics of electromagnetic information in advance and lack of adaptive ability, so as to effectively solve the problem of information leakage detection in electromagnetic signals. The effectiveness of MGCNN in detecting the electromagnetic information leakage of the display is proved through the actual measurement and comparison.

Key words: electromagnetic information leakage electromagnetic radiation Machine learning Convolutional Neural Network information safety computer display TEMPEST