Loading...

Improving Android Malware Detection Techniques

Salehi, Majid | 2016

1645 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 49114 (19)
  4. University: Sharif University of Technology
  5. Department: Computer Engineering
  6. Advisor(s): Kharrazi, Mehdi
  7. Abstract:
  8. Widespread growth in popularity of Android apps stimulates malware authors to consider Android-based devices as an atractive target platform. To defend against this severe increase of Android malwares and help users make a beter evaluation of apps, several approaches have been proposed. However, most of them suffer from some shortcomings such as being computationally expensive, not being general, or not being robust enough.According to obfuscation, encryption, and transformation techniques used in malwares, the static detection techniques are not efficient. Another approach is to use dynamic detection techniques, but existing dynamic techniques suffer from lack of atention to semantic behaviors of applications. In this thesis, we aim to mitigate Android malware spread through providing a robust and lightweight semantic behavior detection system. Considering the limited resources, proposed method performs a behavioral detection for gathering features from applications at run-time. Actually we provide a kernel level behavior tracking technique, then according to extracted behaviors and modeling them we can classify given applications as being malicious or benign.Using real-world Android malwares, we show that the proposed method can clearly reconstruct fine-grained malicious behaviors to ease malware detection. We further evaluate the effectiveness of proposed method with 12026 benign apps from Google Play and 5560 malwares from Drebin and we show that proposed method can detect 92% of malwares with few false alarms
  9. Keywords:
  10. Malwares ; Android Operating System ; Android Applications ; Malware Detection ; Behavior Tracking

 Digital Object List

 Bookmark

...see more