Machine Learning Enhanced Real-Time Intrusion Detection Using Timing Information
Machine Learning Enhanced Real-Time Intrusion Detection Using Timing Information
复制标题
DOI：
--
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
Hang Xu;F. Mueller
中科院分区：
文献类型：
--
作者：
Hang Xu;F. Mueller
过去的工作已经调查了实时控制设备的入侵检测机制。这项工作提供了一种将安全监控和检测与实时控制分离的新框架，其中前者在云边缘设备上执行，而后者在连接到受控系统的嵌入式设备上运行。我们贡献了一个安全监控系统，该系统验证了目标控制器的最坏情况下的时间界限，并通过将其与基于模型的预测（来自机器学习）进行比较来验证其控制输出。
Past work has investigated intrusion detection mechanisms for real-time control devices. This work contributes a novel framework of separating security monitoring and detection from real-time control, where the former is performed on Cloud edge devices while the latter is run on embedded devices attached to the system that is controlled. We contribute a security monitoring system that validates worst-case timing bounds of the target controller and also validates its control outputs by comparing it against model-based predictions, which are derived from machine learning.