AI and the World's Hard Problems · Quantum physics & quantum computing · arXiv · 2026-10-21
Concurrent Entanglement Purification Scheduling in Dynamic Quantum Networks
Quantum networks must deliver end-to-end entanglement whose fidelity meets application requirements, typically through purification along multi-hop paths, but existing schedulers work at the link level or for static request sets and ignore concurrent requests competing for link capacity and quantum memories. The paper studies concurrent entanglement purification scheduling in a time-slotted network with heterogeneous links, formulating a MILP scheduler that maximizes scheduled requests with max-min path allocation as the fairness criterion, plus reinforcement learning schedulers based on DQN, PPO and SAC. Under varying loads, topologies and fidelity targets, the RL schedulers achieved the highest success rates, improving on MILP by up to 6% and on Progressive Filling and Greedy heuristics by 16.5% and 19.5%, at the cost of higher Bell-pair consumption and some request deferral, while MILP gave the highest fairness and lowest Bell-pair consumption in most scenarios.