Towards a Network-based Approach for Smartphone Security

Smartphones have become an important utility that affects many aspects of our daily life. Due to their large dissemination and the tasks that are performed with them, they have also become a valuable target for criminals. Their specific capabilities and the way they are used introduce new threats in terms of information security. The research field of smartphone security has gained a lot of momentum in the past eight years. Approaches that have been presented so far focus on investigating design flaws of smartphone operating systems as well as their potential misuse by an adversary. Countermeasures are often realized based upon extensions made to the operating system itself, following a host-based design approach. However, there is a lack of network-based mechanisms that allow a secure integration of smartphones into existing IT infrastructures. This topic is especially relevant for companies whose employees use smartphones for business tasks. This thesis presents a novel, network-based approach for smartphone security called CADS: Context-related Signature and Anomaly Detection for Smartphones. It allows to determine the security status of smartphones by analyzing three aspects: (1) their current configuration in terms of installed software and available hardware, (2) their behavior and (3) the context they are currently used in. Depending on the determined security status, enforcement actions can be defined in order to allow or to deny access to services provided by the respective IT infrastructure. The approach is based upon the distributed collection and central analysis of data about smartphones. In contrast to other approaches, it explicitly supports to leverage existing security services both for analysis and enforcement purposes. A proof of concept is implemented based upon the IF-MAP protocol for network security and the Google Android platform. An evaluation verifies (1) that the CADS approach is able to detect so-called sensor sniffing attacks and (2) that reactions can be triggered based on detection results to counter ongoing attacks. Furthermore, it is demonstrated that the functionality of an existing, host-based approach that relies on modifications of the Android smartphone platform can be mimicked by the CADS approach. The advantage of CADS is that it does not need any modifications of the Android platform itself