From Provable to Practical: A Problem-Driven Survey of Classical and Machine-Learning Defenses for DV/CV Quantum Key Distribution

Abstract

Quantum key distribution (QKD) promises information-theoretic security, yet practical deployments in discrete-variable (DV) and continuous-variable (CV) settings remain exposed to device imperfections, channel manipulation, finite-key effects, and vulnerabilities in machine-learning (ML) components used for adaptation and monitoring. This survey adopts a problem-driven perspective based on nine practical problem classes (P1-P9) spanning device, channel, protocol, ML, and network layers. For each class, we compare classical defenses with ML-enabled solutions including anomaly detection, parameter prediction, noise estimation, adversarial purification, and resource allocation. Reported results include DBSCAN-based CV attack detection at P=99.7%, R=99.8%, F1=0.998, adversarial robustness recovery up to 79.5%, channel-amplification detection at 100%/91.26% under low/high-noise conditions, and LightGBM-based noise prediction reducing evaluation time by up to 98.8%. The survey further proposes a benchmarking framework combining datasets, stress protocols, and unified evaluation metrics including SKR impact, maximum distance, latency, and robustness. Finally, we provide defense-in-depth deployment guidelines and outline future research directions for secure and practical QKD systems.