Real-time detection of cloud tenant malicious behavior based on CNN
Hao Chen, Ruizhi Xiao, Shuyuan Jin · 2020
Cloud computing provides convenient on-demand services to cloud tenants while bringing many difficulties to cloud providers in detecting tenants' illegal activities. This paper proposes a convolutional neural network (CNN) based approach to detect cloud tenant malicious behaviors in real time. The proposed approach constructs a CNN model to automatically learn sequence patterns from system call sequences and detect tenant malicious behaviors effectively. It further utilizes Spark Streaming techniques, resulting in its capabilities of processing large amounts of tenant data in clouds in real time. The experimental results show that the proposed approach can not only outperform three existing methods but also achieve high detection rates in real time when deployed on the popular cloud platform OpenStack.