Date of Award

3-14-2014

Document Type

Thesis

Degree Name

Master of Science

Department

Department of Electrical and Computer Engineering

First Advisor

Thomas E. Dube, PhD.

Abstract

Every year, malefactors continue to target the Android operating system. Malware which root the device pose the greatest threat to users. The attacker could steal stored passwords and contact lists or gain remote control of the phone. Android users require a system to detect the operation of malware trying to root the phone. This research aims to detect the Exploid, RageAgainstTheCage, and Gingerbreak exploits on Android operating systems. Reverse-engineering 21 malware samples lead to the discovery of two critical paths in the Android Linux kernel, wherein attackers can use malware to root the system. By placing sensors inside the critical paths, the research detected all 379 malware samples trying the root the system. Moreover, the experiment tested 16,577 benign applications from the Official Android Market and third party Chinese markets which triggered zero false positive results. Unlike static signature detection at the application level, this research provides dynamic detection at the kernel level. The sensors reside in-line with the kernel's source code, monitoring network sockets and process creation. Additionally, the research demonstrates the steps required to reverse engineer Android malware in order to discover future critical paths. Using the kernel resources, the two sensors demonstrate efficient asymptotic time and space real-world monitoring. Furthermore, the sensors are immune to obfuscation techniques such as repackaging.

AFIT Designator

AFIT-ENG-14-M-08

DTIC Accession Number

ADA600990

Share

COinS