Analysis Of Sensitive Data Leakage And Sensitive Behaviors In Android Applications

The Android system is currently the most widely used operating system for mobile smart terminals.The analysis and detection related to security on Android applications is a hot issue of current research.This thesis focuses on the approaches and tools for analyzing the sensitive data leakage and sensitive behaviors in Android applications.The current tools for analyzing sensitive data leakage mainly focus on the accuracy with few of them addressing the importance of the efficiency.Actually,the high cost may cause problems such as timeouts or program crashes when the tools analyze large-scale applications,which block them from wide usage.Upon the study of the static analysis of Android applications,this thesis proposes a reasonable and accurate model for Android applications,and designs an efficient static taint analysis approach for detecting sensitive data leakage.In addition,we study the misuse of the runtime permission model,and propose a static tool for detecting the bugs and implicit malicious behaviors in applications.Specifically,the main work and innovations of this thesis are as follows:1.This thesis designs a model of Android applications called statement-tree model,which involves in the information of the classes,methods and statements existed in an app.This model can be utilized in the light-weight analyses.Besides,we propose the concept of the whole call graph(WCG)to simulate the running process of Android applications.The WCG adds the dummy main methods and ICC helper methods into the original call graph and is constructed based on the FlowDroid tool.2.This thesis gives a model for the process of propagation and leakage of the sensitive data.Based on the hypothesis that a taint flow contains a sequence of taint values,we propose taint value graph(TVG)and give the formal definition and properties.To construct the TVG,we define 19 taint rules according to the ways that the taint values propagate in the program.3.This thesis proposes an efficient taint analysis approach FastDroid based on the TVG,which has high accuracy as an object-sensitive,field-sensitive,flow-sensitive and contextsensitive analysis.FastDroid implements a lightweight analysis on the statement-tree model with multiple iterations way to construct the TVGs.To improve the efficiency,we design algorithm to detect multiple taint flows simultaneously.To improve the accuracy,the potential taint flows are checked on the control flow graph to obtain the correct taint flows and eliminated the spurious taint flows.Experiments show that the precision and recall rates of FastDroid are higher than the state-of-the-art tool FlowDroid.Also,FastDroid is significantly better than FlowDroid in efficiency,with less analysis time and higher stability.4.The introduction of the runtime permission model in Android 6.0 greatly strengthens the security of Android applications,but there are still security vulnerabilities.We study the implicit malicious behaviors and bugs caused by the misuse of runtime permissions and give their formal definitions.The user-aware context is proposed to express the relationship of the sensitive behaviors and their driven events.Then,the detection tool RTPDroid is implemented.Experiments show that RTPDroid can effectively detect the implicit malicious behaviors and bugs in applications.In addition,there exist many such problems in real applications.