Research

  1. Xiao Zhang, Yousra Aafer, Kailiang Ying and Wenliang Du. Hey, You, Get Off of My Image: Detecting Data Residue in Android Images. To appear in Proceedings of the 21st European Symposium on Research in Computer Security (ESORICS’16). Heraklion, Crete, Greece. September 26-30, 2016. [acceptance ratio: 60/285 ≈ 21%]
  2. Yousra Aafer, Xiao Zhang, and Wenliang Du. Harvesting Inconsistent Security Configurations in Custom Android ROMs via Differential Analysis. To appear in the 25th USENIX Security Symposium (USENIX Security’16), Austin, Texas, USA. August 10-12, 2016. (Bib) [acceptance ratio: 72/463 ≈ 15.6%]

    Android customization offers substantially different experiences and rich functionalities to users. Every party in the customization chain, such as vendors and carriers, modify the OS and the pre-installed apps to tailor their devices for a variety of models, regions, and custom services. However, these modifications do not come at no cost. Several existing studies demonstrate that modifying security configurations during the customization brings in critical security vulnerabilities. Albeit these serious consequences, little has been done to systematically study how Android customization can lead to security problems, and how severe the situation is. In this work, we systematically identified security features that, if altered during the customization, can introduce potential risks. We conducted a large scale differential analysis on 591 custom images to detect inconsistent security features. Our results show that these discrepancies are indeed prevalent among our collected images. We have further identified several risky patterns that warrant further investigation. We have designed attacks on real devices and confirmed that these inconsistencies can indeed lead to actual security breaches.


  3. Xiao Zhang, Kailiang Ying, Yousra Aafer, Zhenshen Qiu, and Wenliang Du. Life afresidue_esorics2016ter App Uninstallation: Are the Data Still Alive? Data Residue Attacks on Android. In Proceedings of the Network and Distributed System Security Symposium (NDSS), San Diego, California, USA. February 21-24, 2016. (Bib) (Data Residue Vulnerability Website) [acceptance ratio: 60/389 ≈ 15.4%]

    Uninstalling apps from mobile devices is among the most common user practices on smartphones. It may sound trivial, but the entire process involves multiple system components coordinating to remove the data belonging to the uninstalled app. Despite its frequency and complexity, little has been done to understand the security risks in the app’s uninstallation process. In this project, we conduct a systematic analysis of Android’s data cleanup mechanism during the app uninstallation process. Our analysis reveals that data residues are pervasive in the system after apps are uninstalled. For each identified data residue instance, we have formulated hypotheses and designed experiments to see whether it can be exploited to compromise the system security. The results are surprising: we have found 12 instances of vulnerabilities caused by data residues. By exploiting them, adversaries can steal user’s online-account credentials, access other app’s private data, escalate privileges, eavesdrop on user’s keystrokes, etc. We call these attacks the data residue attacks. To evaluate the real-world impact of the attacks, we have conducted an analysis on the top 100 apps in each of the 27 categories from GooglePlay. The result shows that a large portion of the apps can be the target of the data residue attacks. We further evaluate the effectiveness of popular app markets (GooglePlay, Amazon appstore, and Samsung appstore) in preventing our attacking apps from reaching their markets. Moreover, we study the data residue attacks on 10 devices from different vendors to see how vendor customization can affect our attacks. Google has acknowledged all our findings, and is working with us to get the problems fixed.


  4. [Yousra Aafer, Nan Zhang] co-first author, Zhongwen Zhang, Xiao Zhang, Kai Chen, XiaoFeng Wang, Xiaoyong Zhou, Wenliang Du, and Michael Grace. Hare Hunting in the Wild Android: A Study on the Threat of Hanging Attribute References. In Proceedings of the 22nd ACM Conference on Computer and Communications Security (CCS), Denver, Colorado, USA. October 12-16, 2015. (Bib) [acceptance ratio: 128/660 ≈ 19.3%]

    Android is characterized by the complicated relations among its components and apps, through which one party interacts with the other (e.g., starting its activity) by referring to its attributes like package, activity, service, action names, authorities and permissions. Such relations can be easily compromised during a customization: e.g., when an app is removed to fit an Android version to a new device model, while references to the app remain inside that OS. This conflict between the decentralized, unregulated Android customization process and the interdependency among different Android components and apps leads to the pervasiveness of hanging attribute references (Hares), a type of vulnerabilities never investigated before. In our research, we show that popular Android devices are riddled with such flaws, which often have serious security implications: when an attribute (e.g., a package/authority/action name) is used on a device but the party defining it has been removed, a malicious app can fill the gap to acquire critical system capabilities, by simply disguising as the owner of the attribute.

    More specifically, we discovered in our research that on various Android devices, the malware can exploit their Hares to steal the user’s voice notes, control the screen unlock process, replace Google Email’s account settings activity and collect or even modify the user’s contact without proper permissions. We further designed and implemented Harehunter, a new tool for automatic detection of Hares by comparing attributes defined with those used, and analyzing the references to undefined attributes to determine whether they have been protected (e.g., by signature checking). On the factory images for 97 most popular Android devices, Harehunter discovered 21557 likely Hare flaws, demonstrating the significant impacts of the problem. To mitigate the hazards, we further developed an app for detecting the attempts to exploit Hares on different devices and provide the guidance for avoiding this pitfall when building future systems.


  5. Xiao Zhang and Wenliang Du. Attacks on Android Clipboard. In Proceedings of the 11th Conference on Detection of Intrusions and Malware & Vulnerability Assessment (DIMVA), Egham, UK. July 10-11, 2014. (Bib) [acceptance ratio: 14/60 ≈ 23.3%]
    In this paper, we perform a thorough study on the risks imposed by the globally accessible Android Clipboard. Based on the risk assessment, we formulate a series of attacks and categorize them into two groups, i.e., manipulation and stealing. Clipboard data manipulation may lead to common code injection attacks, like JavaScript injection and command injection. Furthermore, it can also cause phishing attacks, including web phishing and app phishing. Data stealing happens when sensitive data copied into the clipboard is accessed by malicious applications. For each category of attack, we analyze a large number of candidate apps and show multiple case studies to demonstrate its feasibility. Also, our app analysis process is formulated to benefit future app development and vulnerability detection. After a comprehensive exposure of the risk, we briefly discuss some potential solutions.

  6. Xiao Zhang, Amit Ahlawat, and Wenliang Du. AFrame: Isolating Advertisements from Mobile Applications in Android. In Proceedings of the 29th Annual Computer Security Applications Conference (ACSAC), New Orleans, Louisiana, USA. December 9-13, 2013. (Extended version) (Bib) [acceptance ratio: 44/231 ≈ 19%]

    Android uses a permission-based security model to restrict applications from accessing private data and privileged resources. However, the permissions are assigned at the application level, so even untrusted third-party libraries, such as advertisement, once incorporated, can share the same privileges as the entire application, leading to over-privileged problems.

    We present AFrame, a developer friendly method to isolate untrusted third-party code from the host applications. The isolation achieved by AFrame covers not only the process/permission isolation, but also the display and input isolation. Our AFrame framework is implemented through a minimal change to the existing Android code base; our evaluation results demonstrate that it is effective in isolating the privileges of untrusted third-party code from applications with reasonable performance overhead.