Exact Resource Laws for Passive Wavelength Routing in Entanglement Networks

The gist The exact resource laws for passive wavelength routing in entanglement networks provide a general way to determine how spectral use, receiver load, repeated connections, and fan-out constrain one another by treating wavelength assignment as a resource-optimization problem

How it works

The paper addresses the challenge of determining the constraints on spectral use, receiver load, repeated connections, and fan-out in entanglement networks by using the network’s connectivity graph where wavelength assignment becomes a resource-optimization problem. For one-sided fan-out, assigning each link to a center and grouping links with the same center gives an exact optimization for arbitrary networks and fan-out limits. When both conjugate wavelengths are allowed to fan out, a balanced binary hierarchy attains the minimum spectral-layer count for a complete network while reducing the maximum receiver load to logarithmic in the number of users. The eight-user complete network explicitly shows the competing roles of spectral efficiency, receiver load, redundancy, and fan-out.

Optical Layers and Graph Representation

The paper defines a spectral layer as a pair of conjugate wavelength channels routed to disjoint user sets Al and Bl, realizing the biclique Bic(Al, Bl) =

5. The realized edge set is the biclique Bic(Al, Bl) =

6. This family of layer graphs is denoted by P =

7. The assignment covers G when every requested edge appears at least once, mP (e) ≥ 1 (e ∈ E), and no layer produces an edge outside E. The wavelength load at user u is χu(P) =

8. Thus, χu(P) counts the spectral layers, and therefore the wavelength channels, received by user u, while χ(P) is the load of the most heavily used receiver. The equivalence between wavelength and graph descriptions is established through Proposition 1 (Optical–graph equivalence), which shows that C(Q) = Q TRconjQ counts conjugate-channel routes between users.

One-Sided Architectures

For one-sided fan-out, a nonredundant one-sided assignment is an edgedisjoint star decomposition, E(G) =

27. The minimum number of nonempty star layers with fan-out at most r is Lmin(G, r) = ΦG(r). Corollary 1 states that the minimum layer count from graph orientations is Lmin(G, r) = ΦG(r). The minimum number of spectral layers with fan-out at most r is Lmin(G, r) = ΦG(r).

Two-Sided Fan-Out and Constraints

When both conjugate wavelengths are allowed to split, the paper compares the minimum layer count and the minimum receiver load for complete graphs. Theorem 5 states that bp(KN) = N − 1, lbc(KN) = lbp(KN) = ⌈log2 N⌉. The hierarchical biclique architecture attains both bp(KN) and lbp(KN). Theorem 7 provides a block-hierarchy construction for bounded two-sided fan-out, showing L = g squared + N − g, χ ≤ g − 1 + ⌈log2 r⌉.

Pair Flux and Pairwise Key Rates

The paper moves to achievable key rates by defining the least supporting pre-protocol pair flux ϕe(R) = F← e(R) = inf[x ≥ 0: Fe(x) ≥ R]. The required pair-generation rate for a star S is ΨS(R) = min s∈SS max e∈S ϕe(Re)se, where a positive numerator divided by zero is +∞. Theorem 9 states that under Definition 1, the simultaneous rate region of a fixed star decomposition is RD(Btot) = (R ≥ 0: X S∈D ΨS(R) ≤ Btot).

Common Broadband Source Model

Under the common broadband source model, the occupied layer rates are then Bl(P, π) = P bπ(l), where P ≥ 0 is a common source-rate scale. Equation (126) is the common-source counterpart of the independent-layer pairgeneration optimization.

Discussion

The paper concludes that passive wavelength sharing should not be judged by spectral-layer count alone, as fewer layers can require higher receiver wavelength load, larger fan-out, repeated links, or more splitter stages. The architecture with the fewest layers or the smallest receiver load need not minimize this source requirement. A spectral saving is useful only when the associated splitter loss and pair-generation requirement still allow the target key rates to be reached. The results are exact at different stages of the analysis, and the graph-theoretic results give exact limits on the wavelength architectures and their resources.

ACKNOWLEDGMENTS

This work was carried out within the project skQCI–Slovak Quantum Communication Infrastructure, funded by the European Union under the Digital Europe Programme, call DIGITAL-2021-QCI-01-DEPLOY-NATIONAL, grant agreement No. 101091548, with 50 % national co-financing from the Recovery and Resilience Plan of Slovakia. The project forms part of the European Quantum Communication Infrastructure (EuroQCI) initiative. Neither the European Union nor the granting authority can be held responsible for them.

Appendix A: Supporting graph-theoretic proofs

The proof of Theorem 4 shows that for even N, removing a perfect matching M and letting CPN = KN − M results in a graph where every user has one excluded partner, and the resulting resources satisfy E(CPN) = N(N − 2) / 2. The proof of Theorem 7 shows that the layer count is g squared + Xg i=1 (ni − 1) = g squared + N − g. The proof of Theorem 8 verifies the five claims in order, showing that the nine-layer partition with side size at most two has the lowest pairgeneration requirement among nine-layer nonredundant profiles with side size at most two. The proof of Theorem 15 compares the closed-form requirements pairwise, showing that Hhier7 ≤ H2×2,rep8 if ηst ≥ q 3 − √5.

Appendix B: Supporting pair-generation proofs

The proof of Theorem 13 shows that the minimum is attained on the compact feasible set by choosing a minimizing cross-layer allocation and an optimizing splitter allocation in every active layer. The proof of Theorem 14 sums the requirement level by level to derive Eq. (105). The proof of Theorem 3 shows that KN has a star decomposition with exactly this profile. The proof of Theorem 5 uses the Graham–Pollak theorem and the Katona–Szemerédi theorem to establish lower bounds for the layer-count and receiver-load constraints. The proof of Theorem 6 verifies that a full binary tree with N leaves satisfies L = N − 1, ∆ = 0, χ = height(T). The proof of Theorem 8 uses the explicit decompositions to show that the nine-layer partition has the lowest pairgeneration requirement among nine-layer nonredundant profiles with side size at most two. The proof of Theorem 5 shows that for complete graphs, bp(KN) = N − 1 and lbc(KN) = lbp(KN) = ⌈log2 N⌉.

1. Resource-Aware Network Design for Quantum Communication Systems: AI systems can optimize network topology by solving exact resource laws for passive wavelength routing, determining how spectral use, receiver load, repeated connections, and fan-out constrain one another based on the network's connectivity graph.

2. Optimal Architecture Selection under Hardware Constraints: AI can select the best entanglement architecture for a given network and hardware limitations by comparing five constrained resource points (like those for K8) against key performance indicators such as minimum spectral layers, maximum receiver load, and required total pair-generation rate.

3. Dynamic Splitter Loss Compensation: The system can adapt to varying splitter losses by calculating the required total pair-generation rate using formulas like Hhier 7 or H2x2 rep 8, ensuring that target key rates are met even when the quality of passive splitting changes.

4. Source Model Adaptation for Broadband Sources: AI can determine the minimum common source-rate scale P star A(Rtar) by comparing different wavelength injections based on a single CW-pumped broadband source spectrum, accounting for coupled accidental coincidences within fan-out layers.

5. Key Rate Estimation Under Coincidence Models: The system can calculate achievable secret-key rates Re using the true and accidental coincidence contributions (Me) derived from the CW model, providing a more realistic assessment of key generation in high fan-out scenarios compared to simple edge-separable models.

Entanglement-based networks provide a scalable framework for multiuser quantum communication by passively routing spectrally correlated photon pairs across interconnected nodes. Several wavelength-allocation schemes have already been demonstrated experimentally, but these designs do not yet provide a general way to determine how spectral use, receiver load, repeated connections, fan-out, and optical loss constrain one another. Motivated by a field-deployed four-node network, we formulate wavelength assignment as an exact resource-optimization problem on the network connectivity graph. We consider uniquely paired wavelength channels and require that no user receive both channels of a pair. For one-sided fan-out, we obtain an exact optimization for arbitrary networks, and closed-form resource laws for complete networks and for networks with one excluded partner per user. Allowing both conjugate wavelengths to fan out changes the resource landscape. In complete networks, a balanced binary hierarchy uses the minimum number of spectral layers among nonredundant assignments while reducing the maximum receiver wavelength load to a quantity that grows logarithmically with the number of users. For eight users, the source-cost ordering depends on splitter transmission: changing the transmission can change which architecture requires less pair generation. We formulate a BBM92 rate model for evaluating the corresponding architectures. For a single continuous-wave broadband source, the occupied wavelength pairs are generated with fixed relative weights, while detector singles and accidental coincidences couple the link rates. The same resource laws can therefore be evaluated using the measured parameters of the hardware that implements the network.