A recent study addresses a key practical problem for future quantum networks: how long an entangled pair should be held in memory before attempting entanglement swapping. The authors analyse two heterogeneous quantum links, equipped with single or multiple memories, that differ in decoherence rates, memory capacities, entanglement attempt rates and success probabilities, as well as classical communication latencies.

The goal is to maximize entanglement swapping throughput under different types of fidelity constraints: strict, probabilistic and mean. To do so, the team develops efficient models that estimate the expected entanglement throughput and the probability density functions of the conditional wait time to swap. With these tools, the maximum holding times for each of the two links are optimized to meet the desired fidelity requirements.

The validity of the proposed model and approach is tested with a discrete-event quantum network simulator, a standard tool for reproducing the behaviour of memory-based quantum repeater architectures. The agreement between the model predictions and the simulation supports its usefulness for designing and tuning this type of system.

The result matters because memory-based quantum repeaters are a central piece of architectures designed to extend entanglement over long distances, and the trade-off between waiting long enough to accumulate useful entanglement and avoiding decoherence degrading fidelity is a critical compromise. Having agile analytical models to set those holding times, rather than relying solely on costly simulations, facilitates the design of networks with very different link characteristics.