A taint analysis framework applied to android applications in the packed state

Abstract

In recent years, Android applications have caused personal privacy leaks frequently. In order to analyze the malicious behavior, taint analysis technology can be used to track the API call chain, build a control-flow graph of function, and determine whether there is a security risk. However, with the continuous escalation of offensive and defensive confrontation of source code, more and more applications use reinforcement technology to prevent security practitioners from performing reverse analysis, therefore it is impossible to analyze function-behavior from the source code. Thus, we design a framework of taint analysis that applied to the Android applications, which automatically unpacks the Android APKs, restores the real source code of the App, performs taint analysis, and generates a control-flow graph of function. Experimental tests showed that the system can cope with the current mainstream reinforcement technology and restore the real Dex file quickly. Simultaneously, compared with the number of nodes before packing, the generated control-flow graph had an explosive increase, which effectively assisted manual analysis of App with the privacy leakage behaviors.