Android malware detection using combination of Support vector machines and fuzzy logic

Nowadays malwares are serious threat for android systems. In recent years with increasing use of Android platform on mobile devices, researchers have focused to this issue more. Various techniques have been introduced for the detection of Android malwares but it seems that growth of these techniques is not comparable with malware growth rate. Every day new malwares hit the Android Market that cannot be identified, and cause serious damage to the hardware and software of mobile devices. The most efficient approach with minimal overhead so far, is using Support Vector Machine (SVM) algorithm. To detect malwares by SVM method, applications are classified in two classes: malware and software. This classification is done by analyzing the features of each program and specific weights which are allocated to features based on the risks that they may have. In this study, a new approach for detecting android’s malwares is proposed. This approach uses fuzzy systems to weight the features and it combines Support Vector Machines and Fuzzy logic. Simulation results show that the proposed approach provides more efficiency and transparency than other methods

Keywords:

Malware , Android , Support Vector Machine , Fuzzy