Mobile cloud forensics: An analysis of seven popular Android apps

Ben Martini, Quang Do, Kim-Kwang Raymond ChooThe Cloud Security Ecosystem 2015 · DOI · DBLP

Abstract

As digital life migrates from local storage to the cloud, forensic investigators face a growing "data gap." This study analyzes seven popular Android applications—including Dropbox, Box, and OneDrive—to determine if local artifacts can provide unauthorized access to remote cloud data. By combining filesystem analysis with memory forensics, researchers successfully gained full cloud access for five out of six primary cloud-syncing apps, revealing significant vulnerabilities in how mobile apps handle authentication tokens and sensitive metadata.

The Forensic Data Gap

Traditional digital forensics often stops at the physical boundaries of a device. However, as users increasingly rely on cloud-syncing services, the most critical evidence frequently resides on remote servers. This creates a challenge for investigators: how to bridge the gap between a suspect's phone and their cloud-stored data. This research systematically evaluates seven apps (Dropbox, Box, OneDrive, ownCloud, Evernote, OneNote, and UPM) to identify forensic artifacts and test the feasibility of secondary cloud access without user passwords.

Methodology: Beyond Standard Imaging

To uncover hidden artifacts, the researchers employed a multi-layered approach:

  • Filesystem Imaging: Creating bit-for-bit copies of device partitions.
  • Signature Bypass: A novel technique involving the modification of the Android services.jar file. By injecting code to bypass signature checks, the researchers allowed custom-signed apps to access the AccountManager API, which typically restricts sensitive data to the original app creator.
  • Memory Forensics: Using HPROF memory heaps and Object Query Language (OQL) to find "volatile" artifacts. This allowed the discovery of session tokens and static API keys that appear in plaintext within the device's RAM during execution, even if obfuscated in the source code.

Forensic Artifact Mapping

The study produced a detailed map of forensic artifacts for each application. Investigators can find a wealth of metadata within SQLite databases and XML shared preference files. For instance, the Dropbox application maintains a comprehensive database of files stored in the cloud, including their paths, sizes, and modification timestamps.

Table schema for the Dropbox 'dropbox' table, detailing file metadata stored locally.

These local databases allow an investigator to reconstruct the entire directory structure of a user's cloud account without ever connecting to the internet, providing a snapshot of the user's digital footprint.

Security Vulnerabilities and Cloud Access

The most significant finding was the ease with which researchers could "impersonate" a device to gain full cloud access.

  • The Power of Tokens: Most apps use OAuth tokens for authentication. While passwords are rarely stored, long-lived "refresh tokens" (often valid for 60+ days) are frequently found in local storage. Recovering these allows for the generation of new access tokens indefinitely.
  • Plaintext Failures: Despite its security focus, ownCloud was found to store user passwords in plaintext within the Android AccountManager and device memory.
  • Encryption Bypasses: In the case of Box, although files on external storage were encrypted, the researchers found the encryption keys and salts stored in the app's private internal storage. With root access, this encryption is effectively nullified.

Beyond file storage, apps like Box maintain detailed logs of user interactions, which can be invaluable for establishing intent or activity timelines.

The BoxEvent table schema, showing how the app logs specific user actions like uploads and previews.

Conclusion

The researchers successfully achieved full account takeover for five out of the six cloud-syncing apps tested. Only Evernote remained partially resistant due to its use of a proprietary, heavily obfuscated binary format ("x-thrift") for authentication. These results demonstrate that for the majority of popular cloud apps, a forensic practitioner with physical access to a device can bridge the data gap and gain unauthorized access to the vast amounts of data stored in the cloud. Future forensic work must continue to evolve alongside app security, focusing on advanced bootloader exploits and the decoding of proprietary binary protocols.

Summary updated

BibTeX
@inproceedings{martini2015mobile,
  author = {Ben Martini and Quang Do and Kim-Kwang Raymond Choo and Ben Martini and Quang Do},
  title = {{Mobile cloud forensics: An analysis of seven popular Android apps}},
  booktitle = {The Cloud Security Ecosystem},
  year = {2015},
  doi = {10.1016/B978-0-12-801595-7.00015-X},
}

Loading related content...

Figure