A Cloud-Assisted Malware Detection Framework for Mobile Devices

Abstract

While mobile applications make our lives more convenient, security concerns may arise when the mobile applications contain malicious code that would harm the mobile devices and their users financially and physically. In this article, we propose a malware detection framework to protect the mobile devices with the help of the cloud, where the cloud is equipped with the facilities for automatic analysis of large amount of new malware generated everyday, and the device is able to detect malicious intents of running software in real-time based on the knowledge of the analyzed malware. We evaluate the performance of our framework with Android-based systems as case studies. In particular, we study the impact of different system configurations on the time required for malware detection, including detecting algorithms (i.e., CNN and SVM), mobile processors (i.e., ARM CPU and NVIDIA GPU), and wireless networks (i.e., Wi-Fi and 3G for the communication between the device and the cloud). To the best of our knowledge, we are not aware of any other work studying performance impacts of the system configurations of the malware detection systems using the physical machines. As the widespread of malware, we believe that our empirical study is useful when designing antivirus software and can be applied to different application domains, such as automotive, and smart home.