Entanglement Swapping in Orbit: a Satellite Quantum Link Case Study

TL;DR

This work analyzes entanglement distribution via a memory-enabled satellite link to connect ground stations, using both a simple analytical model and an event-based QuISP simulation to capture the impact of quantum memory capacity and satellite round-trip latency. It introduces a dual approach: a tractable rate formula r $\le p_{\text{BSM}}\frac{\eta N}{t_{\rightleftharpoons}}$ and a differential-latency correction $r^*$ that accounts for satellite motion, validated against simulations for single and dual-link configurations. The study shows memory size and dynamic memory allocation across links critically shape the achievable entanglement rates, and demonstrates how swapping at the satellite can bridge metropolitan ground networks into a hybrid quantum internet. It also extends QuISP with free-space channels and satellite-aware timing to enable scalable simulations of large satellite–fiber quantum networks, outlining future work on satellite–satellite links and memory-noise effects for practical deployment.

Abstract

Satellite quantum communication is a promising way to build long distance quantum links, making it an essential complement to optical fiber for quantum internetworking beyond metropolitan scales. A satellite point to point optical link differs from the more common fiber links in many ways, both quantitative (higher latency, strong losses) and qualitative (nonconstant parameter values during satellite passage, intermittency of the link, impossibility to set repeaters between the satellite and the ground station). We study here the performance of a quantum link between two ground stations, using a quantum-memory-equipped satellite as a quantum repeater. In contrast with quantum key distribution satellite links, the number of available quantum memory slots m, together with the unavoidable round-trip communication latency t of at least a few milliseconds, severely reduces the effective average repetition rate to m/t -- at most a few kilohertz for foreseeable quantum memories. Our study uses two approaches, which validate each other: 1) a simple analytical model of the effective rate of the quantum link; 2) an event-based simulation using the open source Quantum Internet Simulation Package (QuISP). The important differences between satellite and fiber links led us to modify QuISP itself. This work paves the way to the study of hybrid satellite- and fiber-based quantum repeater networks interconnecting different metropolitan areas.

Figures (6)

• Figure 1: Parameters of both the Micius--Nice and the Micius--Paris links during the passage of the Micius satellite starting January 2$^{\text{nd}}$, 2023 at 10:00 AM. The distance and elevation are computed using orekitorekit and the atmospheric attenuation with LOWTRANLOWTRAN, in a rural setting with $23 \unit{km}$ visibility and no aerosol at a wavelength of $1550\,\unit{nm}$. The channel attenuation is computed using the previous parameters, using (1) of SatLosses for a satellite-to-ground (downlink) transmission with the parameters of dFdP, assuming telescope diameters of 10 cm on the satellite and a 1 m diameter on the ground. The link is supposed to be available only for elevations above 20°.
• Figure 2: Timing of the first photon train, for a Micius--Nice link with $m_S=100$, extracted from our simulation. The reception of the first packet sets $t=0$, and the repetition rate is 1 MHz. The local packet corresponds to the photon emitted by the ground quantum memory and the remote one to the photon received from the satellite. As shown by the inset, while initially in sync, they progressively desynchronize, the 67$^\text{th}$ being the last one inside the acceptance window. A value of 1.5 ns (default in QuISP) was employed for the acceptance window because it is technologically reasonable and consistent with our other parameters.
• Figure 3: Rate of entanglement generation over a single Micius--Nice link for quantum memory sizes $m_S\in\{10,50,100\}$. The orange shaded area corresponds to the expected $1\sigma$ statistical fluctuations. The simulation datapoints (blue points) for each second are within expected fluctuations.
• Figure 4: Number of generated entangled pairs over a single Micius--Nice link for quantum memory $m_S\in\{10,50,100\}$. The orange shaded area corresponds to the expected $1\sigma$ statistical fluctuations. The simulation (blue curve) is within expected fluctuations.
• Figure 5: Comparison of the theoretical entanglement distribution rate over a swapping Nice--Micius--Paris link with dynamic memory allocation against the theoretical and simulated distribution rates with fixed allocation. The simulation datapoints (blue points) for each second are within expected fluctuations.