arXiv:2610.27332
Orchestrating Key Scarcity: Global Capacity Allocation in Hybrid QKD-PQC Networks
Pablo P. Rejo, Manuel Fernández-Veiga, Ana Fernández-Vilas, Diego Cajaraville-Aboy, Javier Vecino-Peñas, Rebeca P. Díaz-Redondo
Abstract
This article presents a Distributed Key Management System (DKMS) that provisions quantum-safe symmetric encryption keys to applications across a network. To overcome the distance and rate limitations inherent to Quantum Key Distribution (QKD), the architecture adopts a hybrid design: it utilizes multi-hop trusted relays for physical key transport, while applying an end-to-end Post-Quantum Cryptography (PQC) layer to ensure the relayed key material remains strictly protected from intermediate nodes. Because QKD transforms key generation into a scarce physical resource, managing unpredictable application demands poses a critical capacity allocation problem. The core contribution of this work is an SDN-driven control plane that globally orchestrates the network. Rather than relying on vulnerable hop-by-hop capacity reservations, a central controller simultaneously computes optimal filling rates and multipath routing for all per-peer key buffers. This global orchestration ensures that no application is starved and that all key buffers synchronize their filling processes. Evaluated on a Kubernetes-based digital twin across five distinct network topologies, the controller successfully filled all key buffers to capacity in under six minutes. Furthermore, under deliberately excessive demand, the network demonstrated graceful degradation, safely rejecting excess requests through controlled rate-limiting rather than experiencing catastrophic failure.
AI Summary
The paper addresses network-wide allocation of scarce QKD key capacity when applications make bursty requests. It proposes a distributed key-management architecture with per-peer transport-key buffers, local token-bucket admission control, and a centralized SDN controller that jointly selects replenishment rates and multipath routes. Transport keys receive end-to-end ML-KEM-based protection against intermediate relays, alongside hop-by-hop OTP protection using link keys. The controller solves a maximum concurrent multicommodity-flow LP with shared bidirectional link capacities. Its base allocation is $r_k=\delta_k+\lambda(B_k-L_k)$, compensating estimated consumption while filling remaining buffer space. With fixed rates and consumption, initially non-full buffers reach capacity together at $T=1/\lambda$. A second optimization distributes residual capacity; demand-deficit variables permit operation under overload. Evaluation uses a Kubernetes digital twin with emulated QKD links across five 20-node topology realizations, each with 380 ordered-pair buffers. All buffers reached 95% of their 8,192-key targets within 340 seconds; median times were 175–245 seconds. Load tests recorded roughly 2.2–2.3 million attempted round-trips per topology, with 8.92–22.24% rejected through HTTP 429 back-pressure. Three realizations had no key mismatches; RGG had 38, and ER had 100,709, an unresolved anomaly. Stated limitations include second-phase solver failure in roughly 70% of solves, triggering shortest-path fallback without LP-certified link loads; an 8-GB memory limit exceeded at 31 nodes; fixed, uniform link lengths; incomplete automatic link-failure reporting; and unstudied control-loop stability under adversarial demand. Reported load-test throughput also draws on prefilled buffers rather than establishing indefinite sustainability.
Central claim may not hold: The synchronization theorem applies to a frozen-rate model without commodity-specific bonuses, whereas the evaluated controller adds bonuses that generally change individual completion times, and its measurements establish only 95%-capacity crossings rather than simultaneous full-capacity completion. After roughly 70% of solves, the controller combines phase-1 rates with shortest-path fallback routing, for which the LP's link-capacity and end-to-end admission guarantees do not apply. The overload formulation penalizes aggregate demand deficits but does not guarantee fair deficit allocation or positive service for every commodity, leaving starvation freedom unestablished. The overload experiment reports 100,709 mismatched keys in ER, approximately 4.42% of attempts, plus 38 in RGG, so integrity-preserving safe degradation is not established across the evaluated networks. The paper incorrectly infers transition safety from separate feasibility of successive allocations, because asynchronous forwarding and rate updates can temporarily overload links.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.