War against mobile malware with cloud computing and machine learning forces

Today´s smart phones are used for wider range of activities. This extended range of functionalities has also seen the infiltration of new security threats. The malicious parties are using highly stealthy techniques to perform the targeted operations, which are hard to detect by the conventional signature and behavior based approaches. Besides, the limited resources of mobile device are inadequate to perform the computationally extensive malware detection tasks and to sustain the device´s clean status. In this paper, we propose an effective and resource rich detection system which uses certain distinguishing combinations of permissions and intents used by the apps to identify the malware apps. Different machine learning algorithms are investigated for classification of apps into benign or malware types. To the best of our knowledge, this is the first ever work in which both the permissions and intents have been amalgamated for malware detection using cloud computing paradigm. Effectiveness of our approach is verified by testing the real-world malware and benign apps collected from various sources.