TL;DR
This paper presents a novel, forensically sound methodology for collecting and analyzing evidence from Android devices, aiming for device agnosticism. It addresses the growing challenge of extracting data from increasingly complex mobile devices, particularly from third-party applications and cloud services, by proposing a structured, eight-step process grounded in established digital forensics principles.
What Problem Does This Paper Address?
The ubiquity of Android smartphones has made them a critical source of digital evidence in criminal and civil investigations. However, the unique security features of Android, such as sandboxing and encryption, coupled with the vast diversity of device manufacturers and OS customizations (fragmentation), pose significant challenges for forensic practitioners. Existing research often relies on rooting devices, which can compromise data integrity and is not always feasible or advisable. Furthermore, much of the existing research is aging rapidly due to the fast-paced evolution of Android. This paper aims to fill this gap by providing a contemporary, forensically sound, and as device-agnostic as possible methodology for evidence collection and analysis from Android devices, with a particular focus on cloud apps.
Methods
The proposed methodology is built upon Martini and Choo's cloud forensics framework and emphasizes two key characteristics: adherence to forensic soundness principles (minimal handling of the original, accounting for changes) and device agnosticism. The methodology comprises eight steps:
- Identify Device and Preserve Evidence: This initial step involves identifying the device and its Android version, and crucially, placing the device in a radio-suppressed environment (e.g., a Faraday bag) to prevent remote wiping. If unlocked, practitioners should check for encryption and lock codes.
- Collect Evidence (Physical Image): This is a critical step that involves collecting a "bit-for-bit" physical image of the device's
userdatapartition. This is preferred over logical collection as it allows for more thorough analysis, recovery of deleted files, and bypasses passcode locks without requiring root access. The paper details techniques for booting a custom live OS in memory and using ADB forwarding to achieve this. - Examine App Files in Private Storage: This involves analyzing files within each app's private directory, which typically contains
shared_prefs(configuration files), databases, cache, and webview data. - Examine App Files on External Storage: This step focuses on analyzing data stored on external storage (physical SD card or emulated storage), where Android's sandboxing rules are more relaxed.
- Examine App Databases: This involves identifying and analyzing various database formats, with a focus on SQLite databases commonly used by apps.
- Examine and Analyze Accounts Data: This step addresses the collection of authentication credentials stored by apps, often via the OS AccountManager service. The paper details techniques to access this data, even when stored in secure physical storage, which may involve modifying the system partition.
- Analyze Apps: This stage involves deeper analysis of the apps themselves, using a combination of static and dynamic analysis techniques. This includes analyzing memory heaps and decompiling app code to understand how data is created and managed.
- Reporting and Presentation: The final step involves documenting the findings, including cryptographic hashes of collected data to demonstrate integrity, and detailing any changes made to the original evidence.
Datasets
Not discussed in the paper. The methodology is presented as a general framework and is demonstrated using seven popular cloud apps in a separate work [4].
Key Insights and Ideas
- Rooting is Generally Undesirable for Forensics: The paper strongly advocates against rooting devices for forensic collection due to the potential for unknown changes to data integrity and the unreliability of exploits.
- Physical Imaging is Paramount: Obtaining a bit-for-bit physical image of the
userdatapartition is highlighted as crucial for comprehensive analysis, especially for recovering deleted data and bypassing security measures. - Device Agnosticism is a Challenge, but Achievable: While Android fragmentation makes true device agnosticism difficult, the methodology focuses on leveraging underlying Android OS and APIs rather than device-specific exploits.
- Cloud Apps Present Unique Challenges: The paper emphasizes the growing importance of analyzing data from cloud computing apps, which often store significant evidential data remotely.
- Understanding App Behavior is Key: Beyond just collecting data, the methodology stresses the importance of analyzing app code and behavior to understand the context and provenance of the data.
Key Results
The paper presents a conceptual methodology and does not report on quantitative results from a specific dataset. The key "result" is the proposed eight-step methodology itself, designed to be forensically sound and practical for modern Android devices. The authors demonstrate the utility of their methodology using seven popular cloud apps in a related work [4].
Limitations and Future Work
- Device Agnosticism is Practical, Not Absolute: The authors acknowledge that achieving complete device agnosticism is impractical due to OEM customizations. The methodology aims to be as agnostic as possible by focusing on underlying OS features.
- Rooting for Specific Tasks: While generally discouraged, the paper notes that rooting might be necessary for certain advanced tasks, such as collecting volatile memory, if the device is already rooted and unlocked.
- Custom Live OS Development: The paper recommends practitioners develop their own live OS for maximum control and understanding, but acknowledges that using a trusted pre-built live OS may be necessary.
- Future Work: The paper suggests that further research is needed to address the challenges of analyzing data from cloud apps and to develop more device-agnostic techniques. The authors also highlight the need for practitioners to stay updated with evolving Android security features and forensic tools.
Summary updated