Graph Neural Network-Enabled Cyber Threat Detection and Quantum-Safe Security Framework for Intelligent Optical Network Infrastructures

Abstract

The increasing adoption of intelligent optical network infrastructures to support 5G/6G 
communications, cloud computing, edge intelligence, Internet of Things (IoT), and large-scale 
data center interconnections has significantly expanded the cyberattack surface of modern 
communication systems. Conventional security mechanisms based on static rule sets, signature 
matching, and traditional cryptographic techniques are increasingly challenged by 
sophisticated cyber threats, dynamic attack propagation patterns, and the emerging risks posed 
by quantum computing. To address these challenges, this paper proposes a Graph Neural 
Network-Enabled Cyber Threat Detection and Quantum-Safe Security Framework for 
intelligent optical network infrastructures. The proposed framework models optical network 
topologies, traffic flows, and inter-device relationships as dynamic graph structures, enabling 
Graph Neural Networks (GNNs) to capture complex spatial dependencies and identify 
anomalous behaviors associated with cyber-physical attacks. The framework integrates optical 
performance monitoring, graph-based threat intelligence, and real-time attack classification to 
detect threats such as distributed denial-of-service attacks, optical jamming, fiber tapping, 
wavelength hijacking, routing manipulation, and control-plane intrusions. To enhance long
term security resilience, the architecture incorporates quantum-safe protection mechanisms 
based on post-quantum cryptographic algorithms and intelligent key management strategies 
designed to withstand future quantum-enabled adversaries. Furthermore, a graph-driven risk 
propagation model is developed to predict attack diffusion across interconnected optical 
network components, enabling proactive mitigation and adaptive security orchestration. 
Experimental evaluation using realistic optical network datasets and cybersecurity benchmarks 
demonstrates that the proposed framework achieves superior detection accuracy, reduced false 
alarm rates, improved attack localization capability, and enhanced network survivability 
compared with conventional machine learning and deep learning approaches. The integration 
of graph intelligence and quantum-safe security mechanisms establishes a robust foundation 
for secure, scalable, and future-proof optical communication infrastructures. The results 
indicate that graph-based cyber threat analytics combined with quantum-resilient protection 
strategies can significantly strengthen the security posture of next-generation intelligent optical 
networks and support the development of autonomous, trustworthy, and resilient 
communication ecosystems.