Scheduling Entanglement Flows in Multi-channel Quantum Networks

TL;DR

This paper proposes resource allocation strategies for entanglement distribution in multi-channel quantum networks, including classical algorithms and a reinforcement learning approach, evaluated through detailed simulations.

Contribution

It introduces a PPO-based entanglement allocation method and compares it with classical algorithms in a comprehensive quantum network model.

Findings

Dynamic Efficient achieves the lowest request delay.

Success Enhancement increases successful entanglement requests.

PPO-based approach balances capacity utilization, delay, and success rate.

Abstract

This paper addresses resource allocation for entanglement distribution in multi-channel quantum networks. A system model is proposed that integrates a multi-channel quantum network architecture with heterogeneous link characteristics and user-centric entanglement request handling, including queuing and retry mechanisms. Classical allocation methods for assigning channels and quantum processors to generate entanglement between end nodes are implemented, including the Dynamic Efficient algorithm, Static Efficient algorithm, and the Success Enhancement algorithm. In addition, a Proximal Policy Optimization (PPO)-based allocation approach driven by a reward function is proposed. These methods are evaluated through multi-slot simulations using metrics such as request delay, total number of successful entanglement requests, network capacity utilization, and the entanglement request handling rate. The results show that Dynamic Efficient achieves the lowest delay, while Success Enhancement improves the number of successful requests through multipath allocation. The PPO-based method provides the best overall balance by improving capacity utilization and achieving both low delay and a high number of successful entanglement requests.