Detection of Information leakage in cloud

Recent research shows that colluded malware in different VMs sharing physical host may use access latency of a resource as a covert channel to leak critical information. Covert channels employ time characteristics to transmit confidential information to attackers. In this manuscript we have made two important contributions and to the best of our knowledge they are novel. One is to propose a framework for detecting covert channel attack based anomalies in cloud using machine learning . This framework is based on application criticality where the degree of error tolerance will be based on the sensitivity of application. This is a useful feature for cloud based applications as SLAs are trade off between service and cost. There are two modules of the framework, a cluster monitor and another module becomes part of hypervisor for monitoring purposes. Second contribution is we propose an approach to distribute the machine learning technique in such a manner to be able to handle imbalance learning as well as enabling SVM to handle large datasets.