摘要
随机模型是一种典型的有原型攻击,其所构建模板的协方差矩阵和传统模板攻击一样,可能会出现指数运算溢出和协方差矩阵不可逆的问题。此外,它需要控制参考设备多次随机设置明文和密钥,从而捕获能耗数据建立模板,这一过程限制了其使用范围。为了解决此问题,提出基于马氏距离的随机攻击方法,针对参考设备设置随机明文和固定密钥建立模板,并将马氏距离应用到随机模型中用于恢复密钥。实验中对基于Atmel XMEGA128D4微控制器实现的AES密码算法实施攻击。结果表明:使用固定或随机密钥构建模板,具有相同分布的加密中间值,都可恢复正确密钥。对比传统模板攻击和通过相关能量分析,提出的攻击方法能够以更少的痕迹恢复正确密钥,使用约50条痕迹可达100%的成功率,可提升密码硬件系统安全性能分析效率。
Stochastic Model(SM)is a typical profiling attack where in building template,the covariance matrix may become singular and cause exponentiation calculation,which is identical to traditional Template Attacks(TA).In addition,the reference device is fully controlled to set up random plaintexts and keys for many times,and captured power consumption data are used to build templates,which limits the usage of SM and traditional TA.In order to solve these problems,stochastic attack based on Mahalanobis distance is proposed in this paper.By studying the techniques of TA,the template is built on reference device where random plaintexts and fixed keys are used as the input,and Mahalanobis distance(statistical tool)is applied to SM to recover keys.The proposed algorithm is used to attack AES which is implemented on Atmel XMEGA128D4 microcontroller.Attack results indicate that the proposed attack method can verify that fixed or random keys have identical distribution of AES sensitive intermediate values when template is built,which can be used to restore the correct keys.Compared with CPA and traditional TA,the correct key can be recovered by the proposed attack algorithm with less power traces,and 100%of success rate can be reached with about 50 traces.The analyzing efficiency of hardware cryptosystem can be improved.
作者
张顺生
罗玉玲
丘森辉
ZHANG Shunsheng;LUO Yuling;QIU Senhui(School of Electronic Engineering,Guangxi Normal University,Guilin Guangxi 541004,China)
出处
《广西师范大学学报（自然科学版）》
CAS
北大核心
2021年第6期33-43,共11页
Journal of Guangxi Normal University:Natural Science Edition
基金
国家自然科学基金(61801131)
广西高校中青年教师科研基础能力提升项目(2020KY02030)
广西研究生教育创新计划项目(YCSW2020100)。
关键词
旁路攻击
模板攻击
随机模型
马氏距离
相关能量分析
side channel attack
template attack
stochastic model
Mahalanobis distance
correlation power analysis
