AI-Assisted ISP and Chip-Off Forensic Framework for Damaged Android Devices.

Area of Science:

• Digital Forensics
• Computer Science
• Artificial Intelligence

Background:

• Physical damage to smartphones is a major obstacle in mobile forensics, often leading to data loss.
• Conventional logical acquisition methods fail when devices are physically compromised.

Purpose of the Study:

• To develop and evaluate an integrated forensic workflow for data extraction from physically damaged smartphones.
• To address the limitations of conventional methods by combining hardware-level intervention with AI-assisted analysis.

Main Methods:

• The study combined In-System Programming (ISP) and Chip-Off memory extraction techniques.
• An AI-assisted artifact localization and prioritization layer using a 1D-CNN classifier was implemented.
• The workflow was tested on 18 physically damaged Android smartphones.

Main Results:

• Hardware extraction successfully produced verified memory images from all 18 damaged devices.
• The AI classifier achieved an F1-score of 0.88 and ROC-AUC of 0.94 for artifact localization.
• Manual review volume decreased by 78%, expert review time by 63%, and time to first artifact by 65%.

Conclusions:

• The integrated workflow effectively recovers data from physically damaged smartphones, overcoming limitations of standard forensic practices.
• The AI component significantly enhances the efficiency of artifact analysis, reducing expert workload and time.
• The study provides a decision model for ISP vs. Chip-Off selection and validated thermal extraction profiles.