Dissertation Defense – Yang Wang

Time: 10am, Aug 3rd, 2017.
Location: Cramer 129.
Student: Yang Wang.

“Android Ecosystem: Risk Assessment and Malware Detection”
As the most popular mobile operating system, Android has become the primary target of malware attacks. In this dissertation, we addressed the security issues facing by Android ecosystem. Several solutions for risk assessment and malware detection were proposed to mitigate the security threats to Android devices. First, we proposed DroidRisk, a framework for quantitative risk assessment of App permissions and apps. With the quantitative risk information, users can easily understand the security risks of permissions requested by an app as well as the app itself. Next, a visual security cue named DRMeter, was designed to display the risk levels of Android permissions and apps assessed by DroidRisk. The results of the user study demonstrate that DRMeter can effectively improve users’ attention on the security risks of Android permissions and apps and help them decide to install an app or not. We then developed ARPMiner, a framework to automatically extract risk patterns related to malicious activities. Experimental results show that ARPMiner can effectively discover risk patterns associated with behaviors of typical Android malware families. The risk patterns were shown to be useful for improving the performance of malware detection. Finally, we proposed DroidSpotter which uses one-class feature selection and classification for Android malware detection. Without using any malware sample for training, DroidSpotter is capable of defending zero-day malware attacks.

Posted on .