Detecting Malware for Android Platform: An SVM-Based Approach

In recent years, Android has become one of the most popular mobile operating systems because of numerous mobile applications (apps) it provides. However, the malicious Android applications (malware) downloaded from third-party markets have significantly threatened users' security and privacy, and most of them remain undetected due to the lack of efficient and accurate malware detection techniques. In this paper, we study a malware detection scheme for Android platform using an SVM-based approach, which integrates both risky permission combinations and vulnerable API calls and use them as features in the SVM algorithm. To validate the performance of the proposed approach, extensive experiments have been conducted, which show that the proposed malware detection scheme is able to identify malicious Android applications effectively and efficiently.

I. Introduction

The rapid growth of mobile devices and remarkable advances in 4G/5G mobile networking technologies have both inspired and facilitated a plethora of mobile applications (apps). In support of these mobile devices, various mobile operating systems have been developed, such as Android, iOS, and BlackBerry 10, etc. Among these mobile operating systems, Android has become the most popular one because of numerous mobile apps it provides. Unfortunately, smartphones running Android have been increasingly targeted by attackers and infected with malicious apps: according to the mobile threat report released by F-Secure in 2014 Q1 [1], over 95% of malicious apps were distributed on the Android platform. Moreover, as the major sources of Android apps, third-party markets contain a lot of cracked or tampered apps, and there is no sign to indicate whether the apps have been checked for security risks or not in these third-party markets. The lack of security inspection on Android apps intensifies the spreading of malicious apps (malware) on the Android platform. Zhou et al. [2] performed some Android malware analysis by popular security software tools, and the detection rate was only between 20.2% and 79.6%, which clearly indicates an urgent and rapidly growing demand on malware detection for Android applications.