Temporal QBER machine learning for QKD attack detection

A machine learning framework using XGBoost achieves 88.01% accuracy in detecting stealthy eavesdropping attacks in BB84 Quantum Key Distribution systems. The method utilizes 63 physics-informed temporal features to identify attacks that fall below the standard 11% error threshold.