Evaluating Relayed and Switched Quantum Key Distribution (QKD) Network Architectures

We evaluate the performance of two architectures for network-wide quantum key distribution (QKD): Relayed QKD, which relays keys over multi-link QKD paths for non-adjacent nodes, and Switched QKD, which uses optical switches to dynamically connect arbitrary QKD modules to form direct QKD links between them. An advantage of Switched QKD is that it distributes quantum keys end-to-end, whereas Relayed relies on trusted nodes. However, Switched depends on arbitrary matching of QKD modules. We first experimentally evaluate the performance of commercial DV-QKD modules; for each of three vendors we benchmark the performance in standard/matched module pairs and in unmatched pairs to emulate configurations in the Switched QKD network architecture. The analysis reveals that in some cases a notable variation in the generated secret key rate (SKR) between the matched and unmatched pairs is observed. Driven by these experimental findings, we conduct a comprehensive theoretical analysis that evaluates the network-wide performance of the two architectures. Our analysis is based on uniform ring networks, where we derive optimal key management configurations and analytical formulas for the achievable consumed SKR. We compare network performance under varying ring sizes, QKD link losses, QKD receivers' sensitivity and performance penalties of unmatched modules. Our findings indicate that Switched QKD performs better in dense rings (short distances, large node counts), while Relayed QKD is more effective in longer distances and large node counts. Moreover, we confirm that unmatched QKD modules penalties significantly impact the efficiency of Switched QKD architecture.