Repeat-until-success quantum repeaters

We propose a repeat-until-success protocol to improve the performance of probabilistic quantum repeaters. Quantum repeaters rely on passive static linear optics elements and photodetectors to perform Bell-state measurements (BSMs). Conventionally, the success rate of these BSMs cannot exceed 50%, which is an impediment for entanglement swapping between distant quantum memories. Every time that a BSM fails, entanglement needs to be re-distributed between the corresponding memories in the repeater link. The key ingredient in our scheme is a repeatable BSM. Although it too relies only on linear optics and photo-detection, it ideally allows us to repeat every BSM until it succeeds. This, in principle, can turn a probabilistic quantum repeater into a deterministic one. Under realistic conditions, where our measurement devices are lossy, our repeatable BSMs may also fail. However, we show that by using additional threshold detectors, we can improve the entanglement generation rate between one and two orders of magnitude as compared to the probabilistic repeater systems that rely on conventional BSMs. This improvement is sufficient to make the performance of probabilistic quantum repeaters comparable with some of existing proposals for deterministic quantum repeaters.