摘要
云计算环境下存在基于内存总线阻塞的侧信道,恶意用户可利用该侧信道以最低权限窃取客户敏感信息.针对这一问题,本文引入时序差分熵和虚拟机自省技术,提出了一种面向云计算的基于内存总线的侧信道攻击检测方法.该方法不仅可依据内存阻塞时序特征及系统负载状况对系统状态分类,而且实现了系统高危态的精确判定和恶意进程定位.实验结果表明:该方法能准确识别攻击的存在性,并能实现恶意进程的定位.
In cloud environments,the memory bus contention-based covert channel can be leveraged by an unprivileged adversary to effectively steal sensitive customer information in guest virtual machines.To address the problem,we propose a detection mechanism for the memory bus contention-based covert channel.In our approach,system running states are extracted with virtual machine introspection(VMI)and timing difference entropy,then classified by system load and timing features of memory contention.Classification results are utilized to recognize the system's high-risk state,and finally locate the malicious process.Experiments show that our approach is capable of detecting the presence of attacks,while also locating the malicious application as well,providing theoretical and technical support for the trusted cloud environment.
出处
《武汉大学学报(理学版)》
CAS
CSCD
北大核心
2016年第5期418-424,共7页
Journal of Wuhan University:Natural Science Edition
基金
国家自然科学基金(61373169)
国家高技术研究发展(863)计划(2015AA016004)
信息保障技术重点实验室开放基金资助项目(KJ-14-110
KJ-14-101)
关键词
云安全
侧信道攻击
时序差分熵
虚拟机自省
cloud security
covert channel attack
entropy of timing difference
virtual machine introspection