Source-linked AI summary
Network Topologies for QKD Networks
Ori Rottenstreich, Ran Hasson Ruso, Eliahu Cohen
TL;DR
QKD topology design lacks a developed framework connecting network reliability and efficiency requirements to graph structure. The paper proposes graph-based cost functions, evaluates small connected graph families, models loss-aware edge capacity, and gives a construction method for larger graphs. Its supported conclusion is that topology comparison can combine operational cost, robustness, hop distance, and loss-sensitive key capacity.
Problem
The paper addresses how QKD-network reliability and efficiency requirements can be expressed and compared through graph topology.
Method
The paper defines graph-based objectives and cost functions, enumerates connected graph families, models edge capacity as Ke ≈SKR(Le), and composes smaller graphs into larger networks.
Results
The paper identifies graph families and constructions using operational cost, robustness, hop distance, and loss-aware edge capacity, while highlighting SKR sensitivity to added loss.
Takeaways & Limitations
QKD topology comparisons can incorporate both graph structure and physical link loss rather than relying only on hop count.
Abstract
from arXiv · showhide
Quantum key distribution (QKD) is a method for distributing cryptographic keys between remote endpoints, enjoying security based on quantum physics. This paper makes a first step towards studying topologies for QKD networks. QKD networks imply several required characteristics for their reliability and efficiency. We express such properties in terms of graph structure. For the comparison of potential graphs, we describe cost functions that allow us to identify families of graphs that are reliable and efficient. To make the discussion realistic, we summarize representative field-reported QKD performance numbers and illustrate the sensitivity of secret-key rate (SKR) to a few dB of additional loss. Last, we present methods to construct large efficient graphs through connecting smaller graphs.
I. INTRODUCTION
The introduction frames QKD topology design as a graph-optimization problem shaped by robustness, limited path length, trust, and loss sensitivity. It motivates loss-aware cost functions using field-reported performance and identifies topology-design directions for reliable, efficient networks.
- QKD distributes cryptographic keys using security based on quantum physics, but quantum computing threatens traditional key-distribution techniques.
- QKD networks add cryptographic capabilities to traditional datacenter communication, computation, and caching requirements.
- Robustness requires networks to handle node and link failures quickly, including through multiple paths disjoint in intermediate nodes or links.
- Trusted-relay QKD makes path length important because each hop consumes link-key capacity and increases trust assumptions, rather than accumulating quantum noise end-to-end.
- A few dB of additional fiber or hardware insertion loss can materially reduce SKR and can drive it to zero beyond a platform-dependent loss budget.
- The paper proposes cost metrics for QKD overhead and robustness, new topologies balancing both, and extensions beyond graph structure.
II. GRAPH PROPERTIES AND EFFICIENT EVALUATION
The paper studies simple graph properties that can affect QKD-network efficiency.
- The study evaluates simple graph properties that can impact the efficiency of QKD networks.
A. Fundamental Graph Properties
The paper represents QKD networks as connected graphs and evaluates topology using structural, loss-aware, and feasibility measures. These measures capture connectivity, path redundancy, latency, and key-generation capacity.
- A QKD network is modeled as a connected graph G = (V, E) with n = |V | nodes.
- Mean degree, max degree, mean distance, and diameter summarize connectivity, node concentration, and key-distribution latency.Mean distance reflects average latency, while diameter represents worst-case latency.
- Critical nodes are nodes whose removal eliminates graph connectivity, while disjoint-path pairs quantify resilience to single-link failures.Disjoint paths can also support multipath secret sharing in trusted-relay QKD networks.
- Each edge receives loss-dependent capacity Ke = max{0, SKR(Le)}, with fiber and insertion loss determining Le.The approximation ηe = 10^-Le/10 models channel transmittance, and SKR scales approximately linearly with ηe in moderate loss.
- A network is QKD-feasible when its positive-capacity subgraph Gfeas remains connected or preserves connectivity for target node pairs.
- For the five-node graph example, mean degree is 2, mean distance is 1.6, and diameter is 3.Nodes A and B are critical; three node pairs have link-disjoint paths.
B. Practical Evaluation for Small Graphs
For small graphs, the paper reduces topology evaluation to representative graph structures rather than examining every possible labeled graph. The supplied examples enumerate candidate families for five-node networks.
- For n ≤ 8 nodes, exhaustive examination of all potential graphs can be intensive because m = n·(n − 1) / 2 edges yield 2^m graphs.
C D E
The paper groups structurally equivalent graphs into unlabeled families and compares connected families using topology-derived properties. It identifies structural relationships relevant to reliability and illustrates the families for five-node graphs.
- E: Fig. 4 visualizes the five-node families and marks critical nodes in red with dots.
- C: A graph family groups graphs with the same structure when node labels are ignored; members share basic properties such as edge count and degree statistics.
- D: For n = 5, there are 21 connected graph families, enabling comparisons of properties affecting QKD network performance.
- D: Table III organizes the 21 connected five-node families by increasing edge count, from 4 to 10 edges.
- E: Adding edges can make nodes non-critical, while degree-1 nodes are non-critical but cannot belong to pairs with two disjoint paths.
- E: Cycle graphs with n edges have no critical nodes and disjoint paths for every node pair, whereas fewer than n edges cannot provide either property.
C. Cost Functions for Graphs
The paper defines graph-based cost functions combining operational cost, reliability, and connectivity, then extends comparison toward capacity-, loss-, and trust-aware QKD objectives. For small graphs, Families 8, 10, and 21 have the lowest costs, while constructions for larger graphs trade fewer edges against diameter and critical-node effects.
- Cost-function components: Operational cost O counts nodes plus edges, with O = α · n + m; for fixed n, only the edge count changes this cost.The paper typically sets α = 2.
- Cost-function components: Reliability combines the number of non-critical nodes with the number of node pairs having link-disjoint paths, weighted by β.The reliability factor lies in [0, 1], and β = 2 gives greater weight to the non-critical-node ratio.
- Cost-function components: Connectivity combines mean node distance and graph diameter, both measured in hops and scaled by log n, with γ balancing the two terms.The paper typically sets γ = 2, giving mean distance greater weight.
- Comparison of graph families: Low cost indicates low operational cost with high reliability and connectivity; among 21 connected n = 5 families, Families 8, 10, and 21 have the lowest values.Family 3, the path graph, has the highest cost because of low reliability and connectivity, more critical nodes, no disjoint-path pairs, and high distances.
- QKD-aware extensions: More edges increase aggregate key-generation resources, while smaller mean distance improves key-transport efficiency; real edge capacities decay with loss, requiring loss-aware evaluation.The loss-aware model uses edge-dependent capacity Ke ≈ SKR(Le).
- QKD-aware extensions: Capacity-aware objectives optimize key flows under per-edge capacities, while trust-aware objectives penalize relay risk; combining them yields a more realistic topology cost function.The proposed capacity utility uses delivered end-to-end key rates, and trust penalties sum risk scores over intermediate relay nodes.
III. CONCLUSIONS AND FUTURE WORK
The paper connects graph structure to key delivery in trusted-relay QKD networks, ranks small unlabeled connected graph families, and offers composition methods for larger networks. It also introduces loss-dependent edge capacity and identifies weighted routing, trust, and realistic-layout evaluation as future directions.
- Conclusions: The paper proposes operational cost, robustness, disjoint paths, and hop distance as structural objectives combined into cost functions Φ and Ψ.These objectives are used for topology design in trusted-relay QKD networks.
- Conclusions: It enumerates and ranks unlabeled connected graph families for small n, illustrates the analysis for n = 5, and gives a composition method for larger networks.The composition approach connects smaller efficient graphs.
- Conclusions: The paper links graph metrics to QKD physics with edge loss Le and capacity Ke ≈ SKR(Le), highlighting SKR sensitivity to a few dB of additional loss.This provides a lightweight loss-aware model beyond purely structural comparison.
- Future work: Future work will optimize capacity-aware routing and key flows, incorporate trust and multipath key splitting, and benchmark topologies on realistic datacenter layouts.The planned evaluations include heterogeneous edge capacities and measured insertion-loss budgets.