Quantum networks are expected to become a key enabler for interconnecting quantum devices. In contrast to classical communication networks, however, information transfer in quantum networks is usually restricted to short distances due to physical constraints of entanglement distribution. Satellites can extend entanglement distribution over long distances, but routing in such networks is challenging because satellite motion and stochastic link generation create a highly dynamic quantum topology. Existing routing methods often rely on global topology information that quickly becomes outdated due to delays in the classical control plane, while decentralized methods typically act on incomplete local information. We propose SatQNet, a reinforcement learning approach for entanglement routing in satellite-assisted quantum networks that can be decentralized at runtime. Its key innovation is an edge-centric directed line graph neural network that performs local message passing on directed edge embeddings, enabling it to better capture link properties in high-degree and time-varying topologies. By exchanging messages with neighboring repeaters, SatQNet learns a local graph representation at runtime that supports agents in establishing high-fidelity end-to-end entanglements. Trained on random graphs, SatQNet outperforms heuristic and learning-based approaches across diverse settings, including a real-world European backbone topology, and generalizes to unseen topologies without retraining.
Quantum networks are expected to become a key enabler for interconnecting quantum devices. In contrast to classical communication networks, however, information transfer in quantum networks is usually restricted to short distances due to physical constraints of entanglement distribution. Satellites can extend entanglement distribution over long distances, but routing in such networks is challenging because satellite motion and stochastic link generation create a highly dynamic quantum topology. Existing routing methods often rely on global topology information that quickly becomes outdated due to delays in the classical control plane, while decentralized methods typically act on incomplete local information. We propose SatQNet, a reinforcement learning approach for entanglement routing in satellite-assisted quantum networks that can be decentralized at runtime. Its key innovation is an edge-centric directed line graph neural network that performs local message passing on directed edge embeddings, enabling it to better capture link properties in high-degree and time-varying topologies. By exchanging messages with neighboring repeaters, SatQNet learns a local graph representation at runtime that supports agents in establishing high-fidelity end-to-end entanglements. Trained on random graphs, SatQNet outperforms heuristic and learning-based approaches across diverse settings, including a real-world European backbone topology, and generalizes to unseen topologies without retraining.
Quantum networks are becoming increasingly important because of advancements in quantum computing and quantum sensing, such as recent developments in distributed quantum computing and federated quantum machine learning. Routing entanglement in quantum networks poses several fundamental as well as technical challenges, including the high dynamicity of quantum network links and the probabilistic nature of quantum operations. Consequently, designing hand-crafted heuristics is difficult and often leads to suboptimal performance, especially if global network topology information is unavailable.
In this paper, we propose RELiQ, a reinforcement learning-based approach to entanglement routing that only relies on local information and iterative message exchange. Utilizing a graph neural network, RELiQ learns graph representations and avoids overfitting to specific network topologies - a prevalent issue for learning-based approaches. Our approach, trained on random graphs, consistently outperforms existing local information heuristics and learning-based approaches when applied to random and real-world topologies. When compared to global information heuristics, our method achieves similar or superior performance because of its rapid response to topology changes.
Quantum Information networks (QIN) aim to interconnect quantum systems via quantum repeaters through end-to-end entanglement of qubits. The integration of both satellites and ground segments is essential to provide global connectivity, even over large distances. In this paper, we study the problem of entanglement routing in large-scale quantum networks -i.e., with a large number of nodes, with a hybrid space-ground architecture. We decompose the routing problem into two subcomponents: request scheduling and path selection, and structure our approach around two phases: a proactive and a reactive step. We design and implement an entanglement routing strategy in an emulated European quantum network, first in a custom Python-based simulator for performance forecasting, and then into the NetSquid quantum network simulator. Our work thus provides a tool allowing for a comparison between terrestrial and satellite segments in terms of successful end-to-end entanglement in a large-scale network and shows that both segments are required for optimal service. Our results indicate that, within the emulated European topology, terrestrial networks can support end-toend entanglement for approximately 35% of user pairs -the remaining pairs are too distant and suffer from prohibitive losses. In contrast, the satellite-based network achieves connectivity for up to 91% of user pairs, although some pairs remain unreachable due to a lack of simultaneous satellite visibility. However, while satellites can outperform terrestrial networks in total photon distribution, they may not be suitable for lowlatency requests if out of visibility, making ground networks essential. This highlights the complementary roles of space and terrestrial segments in quantum networking.
This paper proposes a novel approach to distributed learning leveraging the dynamic topology of quantum entanglement networks. The core idea is to exploit entanglement-based quantum routing to facilitate direct information exchange between nodes, bypassing the limitations of classical communication bottlenecks. Furthermore, the evolving topology of the network is utilized to optimize the learning process. We introduce a mathematical framework to describe this system, focusing on the key concepts of entanglement distribution, quantum routing, and topological network evolution. The proposed system aims to achieve accelerated learning and overcome the challenges associated with traditional distributed neural networks. The theoretical analysis demonstrates the potential for improved scalability and efficiency in complex learning tasks. The central formula for calculating the entanglement fidelity between nodes is given by *F(s_1, s_2) = 1 - ||s_1 - s_2||/1*, where *s_1* and *s_2* represent the quantum states at nodes 1 and 2, and *||.||* denotes the norm. The learning rate update rule, considering entanglement-assisted information propagation, is defined as *Δw_i = η * (∑_(j∈N(i)) α_ij * (s_i - s_j)) / ||s_i||*, where *w_i* is the weight update for node *i*, *η* is the learning rate, *N(i)* is the set of neighbors of node *i*, *α_ij* is the entanglement fidelity between nodes *i* and *j*, and *s_i* and *s_j* are the quantum states at nodes *i* and *j*. This framework provides a foundation for further research and development in quantum-enhanced distributed learning. ---
With the continuous expansion of quantum communication networks and the increasing diversity of their business types, how to achieve efficient configuration under limited quantum resources has become the key to improving system performance. The quantum resources described in this article mainly include entanglement pairs, quantum bits, and channel capacity. Traditional optimization methods typically rely on precise mathematical models, which are difficult to adapt to dynamically evolving network environments and susceptible to noise interference. Therefore, this study proposes a machine learning based framework for optimizing the allocation of quantum communication resources. This framework integrates deep reinforcement learning and graph neural network technology, enabling collaborative optimization of entanglement allocation, routing selection, and time slice scheduling between nodes. Three typical quantum communication scenarios were constructed and tested in a simulation environment, including static topology, dynamic links, and high noise environments. The proposed method was compared and analyzed with integer linear programming, heuristic greedy algorithms, and rule-based strategies. Experimental data shows that this research method achieved an improvement of approximately 12% to 18% in average fidelity retention, a 15% to 22% increase in resource utilization, and a 20% reduction in latency. This study provides useful reference for the intelligent resource management of large-scale quantum networks in the future.
Entanglement generation in long-distance quantum networks is challenging because resources are limited and entanglement swapping is probabilistic. To maximize the rate of successful requests, existing quantum routing algorithms often rely on computationally expensive methods such as Integer Linear Programming (ILP) to determine which links to entangle and use for end-to-end entanglement generation. However, these approaches fail to meet the latency requirements of real-world quantum networks. In this study, we propose a Reinforcement Learning (RL)-based model that determines which links to entangle in each time slot, replacing the slow ILP-based link-selection phase used in prior algorithms. The proposed Deep Q-learning model is up to 19.2× faster than linear programming in link-selection while maintaining comparable routing performance. The RL link-selection model alone matches ILP in request success rate; combining RL link selection with entanglement caching and proactive swapping raises throughput by up to 52.55% over ILP. Overall, our approach achieves success rates exceeding state-of-the-art solutions while reducing execution time by more than an order of magnitude. Experiments on a synthetic 50-node Waxman topology and the real-world 54-node SURFnet core topology confirm that these gains generalize across network structures, providing a practical path toward scalable quantum routing.
Quantum memories are a critical component of entanglement-based quantum networks, enabling the storage and synchronisation of quantum states across dynamic links. However, current quantum memories have significantly lower capacity than the rate at which entanglement can be generated, making memory saturation a key bottleneck that reduces network efficiency and hinders the scaling of quantum networks. This problem is especially pronounced in dynamic satellite-based quantum networks, where short visibility windows constrain link availability. In this paper, we present a support entanglement-swapping algorithm that utilises leftover entanglement in quantum memories, thereby alleviating memory saturation and increasing network connectivity. Our algorithm combines two mathematical concepts, line graphs and maximum-cardinality matching, to select independent entanglement swap pairs without sharing any entanglement between concurrent swaps. This property ensures that the resulting changes to the network remain local and mutually independent, making the algorithm easy to integrate alongside any existing routing schemes without requiring network-wide coordination. We evaluate the algorithm through simulations on both static fibre-based networks and dynamic satellite networks. Across most configurations, our algorithm increases both the mean and the total number of entanglements shared between end nodes, while also increasing the network's long-range connectivity. The 'SwapWithToUse' algorithm variant consistently provides the greatest improvements, with gains increasing as entanglement-generation rate increases.