CloudSylla: Detecting Suspicious System Calls in the Cloud
Marc Kührer, Johannes Hoffmann, Thorsten Holz
16th International Symposium on Stabilization, Safety, and Security of Distributed Systems (SSS), Paderborn, Germany, September 2014
To protect computer systems against the tremendous number of daily malware threats, security software is typically installed on individual end hosts and the responsibility to keep this software updated is often assigned to (inexperienced) users. A critical drawback of this strategy, especially in enterprise networks, is that a single unprotected client system might lead to severe attacks such as industrial espionage. To overcome this problem, a potential approach is to move the responsibility to utilize the latest detection mechanisms to a centralized, continuously maintained network service to identify suspicious behavior on end hosts and perform adequate actions once a client invokes malicious activities.
In this paper, we propose a security approach called CloudSylla (Cloud-based SYscaLL Analysis) in which we utilize a centralized network service to analyze the clients' activities directly at the API and system call level. This enables, among other advantages, a centralized management of signatures and a unified security policy. To evaluate the applicability of our approach, we implemented prototypes for desktop computers and mobile devices and found this approach to be applicable in practice as no substantial limitations of usability are caused on the client side.[PDF]