基于主成分分析的指令旁路模板分析方法  

Principal Component Analysis in Retrieve Instructions with Side Channel Template

作　　者：吴恒旭[1,2] 陈开颜[1] 邓高明[1] 张鹏[1] 邹程[1] 

机构地区：[1]军械工程学院计算机工程系,河北石家庄050003 [2]成都军区78086部队,四川成都610000

出　　处：《微电子学与计算机》2011年第3期144-148,共5页Microelectronics & Computer

基　　金：国家自然科学基金项目(60940019);军械工程学院原始创新基金(YSCX0903)

摘　　要：提出一种微控制器指令逆向恢复方法.根据微控制器内执行的指令与功耗、电磁等旁路信号之间的相关性,为不同指令分别构建旁路模板,将未知指令的旁路轨迹与模板进行匹配来恢复指令.为解决旁路轨迹中样本点数量较大,导致的模板构建与匹配效率较低的问题,利用主成分分析(PCA)的方法从功耗轨迹中提取旁路信号主成分.实验结果显示,在约400个样本点中选取约30个分量的相同条件下,PCA方法获得的匹配正确率比人工选取有效点方法提高近50%.An method of instruction retrieve is proposed for microcontroller,which can takes the advantage of the relationship between instructions executed in microcontroller and the side channel leakages such as power consumption and electromagnetic emissions,builds side channel templates for instructions and matches the side channel trace from an unknown instruction to retrieve the operation.However,there are usually so many points in a side channel trace that the compute efficiency is very low.An interest component selection method with principle components analysis（PCA） is proposed to select interest components from side channel trace.With about 30 components selected from about 400 sample points,the correct matching ratio gained by PCA is about 50% higher than that by selecting points artificially.

关 键 词：逆向工程 指令恢复 旁路模板分析 主成分分析 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]