作者: Gregory Levitin , Liudong Xing , Yanping Xiang
DOI: 10.1016/J.RESS.2020.106969
关键词: Virtual machine 、 Voting 、 Computer security 、 Optimization problem 、 Cloud computing 、 N-version programming 、 Block (data storage) 、 Warning system 、 Computer science 、 Service (business)
摘要: Abstract Due to the virtual machine co-resident architecture, cloud computing systems are vulnerable attacks (CRAs) where a malicious attacker may access and corrupt information of target user through co-locating their machines on same physical server. To defend against cyber threats such as CRA, early warning mechanisms have been developed with aim detect block an attack at nascent stage. In this paper, we study optimal strategy allocating resources CRAs for voting-based N-version programming (NVP) service running in cloud. A probabilistic model is proposed evaluate failure probability NVP program further expected cost loss considered service. Optimization problems co-determining numbers versions agents solved minimize loss. As demonstrated examples, resultant strategies can effectively allocate defense CRAs.