Demonstration of a reconfigurable quantum network architecture suitable for ground-to-space communication

TL;DR

This work tackles the challenge of integrating satellite links into metropolitan quantum networks by demonstrating a reconfigurable topology that switches between multipoint-to-point entanglement during satellite passes and a ground-based, pairwise uplink when the satellite is unavailable. The approach combines time and frequency multiplexing, using a PPLN-based SPDC source and a frequency-to-time mapping scheme to assign distinct temporal slots to multiple spectral channels, enabling a scalable five-node network with enhanced coincidence-to-accidental ratio (CAR). The authors show that multiplexing yields linear CAR gains and improves two-photon visibility, QBER, and asymptotic secure key rate (SKR) in a quantum key distribution context, with practical considerations for field deployment and potential scalability to 72 users through integrated photonics. The study discusses architectural refinements (flex-grid, passive demultiplexing, on-chip FTM) and outlines a feasible path toward global quantum networking, including robust operation under satellite passes and challenging loss conditions, while highlighting the role of integrated photonics in achieving portable, scalable networked quantum systems.

Abstract

We experimentally demonstrate a reconfigurable quantum network architecture suitable for integrating satellite links in metropolitan quantum networks. The network architecture is designed such that once a satellite is in range, it is configured in a multipoint-to-point topology where all ground nodes establish entanglement with the satellite receiver using time multiplexing to optimize long-distance transmission. Otherwise, the satellite up-link can be rerouted to the ground nodes to form a pair-wise ground network. Leveraging both the time and frequency correlations of our photon-pair source, we demonstrate an increased coincidence-to-accidental ratio without additional resource overhead in a five-node network. To contextualize these experimental findings, we project their performance in a quantum key distribution scenario and outline a feasible route toward field deployment, using integrated photonics to enable network integration of up to 72 users.

Figures (7)

• Figure 1: Experimental implementation of the reconfigurable quantum network for (a) the satellite configuration and (b) the ground configuration. To emphasize the differences between the two configurations, optical elements not used in each configuration are depicted in grey. (SHG: second-harmonic generation, DM: dichroic mirror, HWP: Half-wave plate, BPF: band-pass filter, WG: waveguide, LPF: long-pass filter, Circ: circulator, FBG: fibre Bragg grating, SNSPD: superconducting nanowire single photon detector, Si-APD: silicon avalanche photodetector)
• Figure 2: The quantum‐correlation layer for each switch position (top, middle and bottom) in the signal demultiplexing stage. (a) corresponds to the satellite configuration, while (b) and (c) correspond to the ground configuration.
• Figure 3: The measured spectrum of the signal (a) and idler (b) photons. Wavelength correlated channels $\{\lambda_{i},\Lambda_{i}\}$ are denoted with the same color in both spectrums and reflected by $FBG_{s(i)}$ respectively in Fig. \ref{['fig:setup']}. 1 nm guard bands are used between the signal channels $\{\lambda_i\}$ to minimize cross-talk between the channels.
• Figure 4: Multiplexing scheme performance estimation. In (a), the coincidence-to-accidentals ratio is measured as a function of the pump power. In (b), (c), and (d) the two-photon interference visibility, quantum bit error rate (QBER), and secure key rate are respectively estimated from the measured CAR. For the ground scenario shown in (a), coincidences at low pump power were measured using a SNSPD and APD detector per user, due to the SNSPD's poor detection efficiency at 780 nm. At higher pump power, signal and idler photons were jointly detected on a single SNSPD channel per user. Here, MUX refers to the multiplexed configuration of the network, while No MUX denotes the single-channel baseline.
• Figure 5: Trade-off between the number of frequency-time multiplexed channels and the repetition rate assuming a pump power of $100$$\mu$W, 30 dB channel loss, detector dead time of 1 $\mu$s, a dark-count rate of 1000 counts/s, and a timing jitter of $130$ ps.