Joint Optimization of Routing and Purification to Meet Fidelity Targets in Quantum Networks

TL;DR

A cost-based scheduler that jointly optimizes path selection and purification round, along with two hop-level estimators to predict the minimal purification rounds needed for target hop fidelity, which reduces mean latency and increases success rates compared to fixed-round purification with FIFO scheduling.

Abstract

Quantum networks rely on high-fidelity entanglement links, but achieving target fidelity often increases latency and Bell pair consumption due to purification. This paper proposes a cost-based scheduler that jointly optimizes path selection and purification round, along with two hop-level estimators (a Deep Neural Network classifier and a Bayesian optimizer) to predict the minimal purification rounds needed for target hop fidelity. The scheme flexibly adjusts final entanglement fidelity while minimizing latency, improving request success rates and efficient Bell pair usage. Simulations integrating purification, entanglement generation, and network-level scheduling show that our approach reduces mean latency by up to 8% and increases success rates by 14% compared to fixed-round purification with FIFO scheduling.

Figures (2)

• Figure 1: DNN training structure. The dataset consists of $10,000$ data points per hop, splitting into training and validation sets with an $80/20$ ratio.
• Figure 2: CDF of fidelities across different purification schemes in network load $\lambda = 2$.