Deans

Eric Ackerman, Ph.D. – Graduate School of Computer and Information Sciences

Award Date

1-1-2015

Abstract

As one of the major operating systems adopted by mobile devices, security issues related to Android platform is gaining increasing attention in the research literature in recent years. Android offers users the convenience to install applications through its app stores or third-party sites, but also provide the opportunity for distribution of malicious applications (or malware). Various malware detection methods have been proposed in the literature. These methods can generally be categorized as static analysis or dynamic analysis techniques, with varied degree of success. However this issue is far from addressed. In this research, we propose to tackle the detection of malware in Android platforms using dynamic analysis with a combination of finely tuned machine learning and statistical methods. It has been shown that a combination of machine learning and statistical methods can be effective in detecting attacks in traditional networks. We believe the same philosophy can be applied to this problem area. The new methodology will be compared against peer approaches in terms of a variety of evaluation metrics such as detection rate, false alarm rate, CPU usage, bandwidth consumption, battery consumption, among others. In the long term, it is anticipated that a stand-alone malware detection application will be available for users to download from the app store.

Share

COinS