Quantum key distribution networks (QKDNs) provide a practical approach for delivering quantum-secured key services to existing communication infrastructures. As QKDNs evolve from point-to-point QKD systems toward carrier-grade network infrastructures, standardization becomes essential for defining common architectures, functional models, interfaces, control and management mechanisms, security requirements, and service integration procedures. This paper presents a standardization-oriented review of QKDNs, with emphasis on their architectural foundations, operational functions, deployment challenges, and evolution toward quantum networks. First, we introduce the basic concept of QKDNs and analyze the global standardization ecosystem involving ITU-T, ETSI, ISO/IEC, IETF, IEEE, and related coordination mechanisms. Then, we describe the QKDN framework specified by ITU-T, including the architecture, layered functional model, reference points, key management, control and management, QoS support, resilience, and interworking. Based on this framework, we discuss the major challenges for carrier-grade QKDN deployment, including multi-provider interoperability, cost-efficient integration with existing optical communication infrastructures, coexistence between QKD and classical optical channels at the transmission layer, cross-layer service orchestration, scalability for mobile, low-altitude, and satellite-assisted scenarios, as well as QKDN extension for end-to-end modern cryptography services. Finally, we look at how QKDN standardization may support the broader evolution from trusted-node-based QKD networks toward quantum networks. The objective is to provide a systematic architectural and operational perspective on standardized QKDNs and to outline future research and standardization directions for scalable quantum-secured networking.
Virtualization and softwarization have become critical tools for enhancing the flexibility, scalability, and cost efficiency of optical access networks. While significant progress has been made in virtualizing upper-layer functions of optical line terminals (OLTs), the softwarization of lower-layer functions, particularly physical-layer (PHY) processing, remains a major challenge due to the stringent real-time and high-throughput requirements. This paper provides a comprehensive overview of recent advancements in virtualization and PHY softwarization in optical access systems and presents an experimental study on the scalability of graphics processing unit (GPU)-based virtualized OLT PHY processing. We implement a software-based OLT physical coding sublayer (PCS) on multiple GPU platforms and evaluate the system performance under increasing numbers of consolidated OLT instances. Through detailed measurements of throughput, GPU utilization, memory bandwidth utilization, and CPU–GPU interconnect bandwidth, we investigate system-level constraints impacting scalability. The results show that throughput scales as the number of OLT PCS instances reaches approximately 4–5, but saturates when GPU resource utilization reaches 100%. In contrast, memory and interconnect bandwidth remain underutilized, suggesting that computational resources are the dominant limiting factor. These findings provide important insights into the design of scalable softwarized OLT systems and highlight the potential of leveraging commodity GPU platforms for efficient deployment of future optical access networks.
