AI and the World's Hard Problems · Quantum physics & quantum computing · arXiv · 2026-10-20
Simulation-Based Quantum Forensics for QKD: Machine Learning Attack Detection and Signature Extraction
Practical quantum key distribution (QKD) systems are exposed to implementation-level attacks on optical channels, detectors, and controller behavior. The authors propose QF-LOG, a simulation-based framework that analyzes QKD operational logs to classify normal operation and three simulated attack conditions. Learning models using all four features reached macro F1-scores of 0.8211 to 0.8243, but dropped to 0.5645 to 0.5669 on a dataset generated under shifted simulated operating conditions.