摘要
针对FPGA在运算时会产生电磁泄露的现象,以简单逻辑运算为攻击目标,提出了一种基于支持向量机(SVM)的旁路攻击方法。通过电磁泄露信息采集平台实现电磁泄露信息的采集,以电磁泄露信息为特征信号,通过独立成分分析(ICA)对其进行特征提取。将经过ICA处理之后的信号作为新的特征向量,用已知逻辑运算情况下的样本训练多分类SVM,用训练好的多分类SVM对未知的简单逻辑运算的电磁泄漏信号进行分类。实验结果表明该方法能成功识别加、减、异或、左移、右移运算,其准确率达到了91.4%。
Aimed at the electromagnetic leakage while FPGA is executing,in this paper,the simple logic operations are considered as the attack target,and a kind of side channel analysis method is proposed based on support vector machine（SVM）.Electromagnetic leakage information is captured through electromagnetic leakage information capture platform.Regard electromagnetic leakage information as characteristic signal,feature extract is accomplished by ICA which is considered as the new feature vectors.Then they are used to train by multi-class SVM combined with known classification logic operation.After that a sample of electromagnetic signal with unknown classification logic operation is classified with the trained data.The experimental results show that this method could successfully recognize addition,subtraction,exclusive OR,left shift and right shift,and the accuracy rate reaches 91.4%.
作者
米芳
张洪欣
MI Fang ZHANG Hong-xin(School of Electronic Engineering, Beijing University of Posts and Telecommunications,Beijing 100876,China)
出处
《滨州学院学报》
2017年第2期66-69,共4页
Journal of Binzhou University
基金
国家自然科学基金资助项目(61202399
61571063
61472357)
