A virtual machine anomaly detection system for cloud computing infrastructure

In recent years, cloud computing is used widely due to the vital growth of cloud resources along with its great performance, effective use, and on-demand resources availability. On the other hand, the cloud is uncertain due to various intrusion attacks such as malicious software and backdoors creation. The virtualization concept is popular in the cloud due to its virtual concept for hardware and software. In the cloud, the employment of an intrusion detection system (IDS) needs a scalable and virtualized infrastructure for reducing the traffic that is often carried by the virtualized server. In this work, the cloud platform-based virtual machine anomaly detection system is designed. The Globality-aware Locality Preserving Projection and Incremental Virtual Machine Workload Clustering Algorithm are designed for important modules including data dimension reduction and anomaly detection in the system. From the experimental analysis, it is inferred that the design not only improves accurateness of the online anomaly detecting mechanism, but also the average operation time is less than 0.25s consistent with the timeliness requirements, shortens the average calculating time to a greater extent, improves the timeliness, and achieves the objective of anomaly detection.