A resource-centric, task-based approach to quantum network control

TL;DR

Quantum networks require new control-plane principles beyond classical layered stacks due to non-clonability, fragile entanglement, and short qubit lifetimes. The authors propose a resource-centric, task-based framework where applications specify objectives and network resource managers synthesize sagas—distributed task sequences operating on channels, entanglement, and classical messaging, with topology information broadcast to maintain a unified view. They formalize core constructs (objectives, resources, tasks, capabilities, sagas) and analyze how topology, locking, and execution modes (orchestration vs choreography) influence performance and reliability, including graph-state structures $G=(V,E)$. This approach aims to enable modular, interoperable, and low-overhead quantum-network control that better handles dynamic entanglement and heterogeneous hardware, providing a foundation for future standardization and benchmarking. The framework can be implemented in centralized or distributed forms, offering flexibility in saga derivation and execution.

Abstract

Quantum networks exhibit fundamental differences from their classical counterparts. These differences necessitate novel principles when organizing, managing, and operating them. Here we propose an unconventional approach to organize and manage the operations of quantum network devices. Instead of a hierarchical scheme using layers, like in classical networks and present quantum network stack models, we propose a resource-centric task-based scheme. In this scheme, quantum applications pose objectives, initiated by a node, to a quantum network, such as sharing an entangled state or sending a qubit along a path. The quantum network node initiating the objective consequently derives a distributed workflow, referred to as saga, comprising numerous tasks operating on resources, which completes the objective. We identify three different kinds of resources with their own and independent topology, namely classical messaging, quantum channels and entanglement. Sagas can either be centrally orchestrated or performed in choreography by the network nodes. The tasks of a saga originate from and operate on resources of the network, such as quantum channels or entanglement, and they not only comprise operations and measurements, but potentially also include other tasks or even entire protocols, such as sending a qubit, distributing entanglement or performing entanglement purification steps.

Figures (10)

• Figure 1: The standard OSI model for classical computer networks. It comprises in total seven, hierarchically organized, layers. Data passes from layer to layer in terms of a packet. Each layer adds additional information specific for its layer to the packet, referred to as header. Each layer only operates on the header information belonging to its layer.
• Figure 2: The figure depicts the proposed quantum network control framework. Quantum applications create objectives, like for example distributing a Bell-state. The network resource manager of a node uses its knowledge about the resources of the network node (channels, entanglement, classical messaging) to compute a saga, consisting of distributed tasks and protocols, to achieve the objective. The quantum network nodes consequently execute the saga to achieve the objective.
• Figure 3: Every quantum network node has a global view on all the resources available in a quantum network. Classical messaging changes this view, for example when new entanglement was established, or channels become available.
• Figure 4: The figure depicts the tasks that quantum network node 1 offers. The node offers five entanglement tasks, like entanglement-swap or merging graph states. For channel resources it can send qubits. Lastly, node 1 can implement arbitrary operations (including also quantum error correction).
• Figure 5: The Midpoint task/protocol can be decomposed of several other elementary tasks. In fact, it comprises two Bell-state preparation tasks followed by two synchronized send tasks and an entanglement swapping step in the central node.