GHz-rate polarization-based QKD system for fiber and satellite applications

TL;DR

This work addresses the challenge of delivering secure quantum keys at high rates over both fiber and satellite-like links. It introduces a GHz-rate, polarization-encoded DV-QKD system based on the efficient BB84 protocol with iPOGNAC modulators, achieving up to $R = $ $1.5$ GHz and an intrinsic $QBER$ around $0.4\%$. The authors demonstrate the setup over a laboratory fiber and a 620 m intermodal free-space path in daylight, achieving a sustained SKR exceeding $1$ Mb/s and, under finite-size security bounds, positive SKR up to channel losses of $52$ dB in the lab and $38.5$ dB for satellite-like conditions (≈ $6.5$ kb/s with $N=10^4$). The results, enabled by Qubit4Sync synchronization and turbulence compensation via a fast steering mirror, indicate strong potential for metropolitan quantum networks and space-based QKD, with future work targeting space-qualified hardware for full satellite demonstrations.

Abstract

Quantum key distribution (QKD) leverages the principles of quantum mechanics to exchange a secret key between two parties. Despite its promising features, QKD also faces several practical challenges such as transmission loss, noise in quantum channels and finite key size effects. Addressing these issues is crucial for the large-scale deployment of QKD in fiber and satellite networks. In this paper, we present a 1550 nm QKD system realizing the efficient-BB84 protocol and based on the iPOGNAC scheme. The system achieved repetition rates up to 1.5~GHz and showed an intrinsic QBER of $\sim 0.4\%$. The system was first tested on a laboratory fiber link and then on an intermodal link in the field, consisting of both deployed fiber and a 620 m free-space channel. The experiment was performed in daylight conditions, exploiting the Qubit4Sync synchronization protocol. With this trial, we achieved a new benchmark for free-space BB84 QKD systems by generating a sustained secret key rate (SKR) above 1~Mb/s for 1 hour. Finally, exploiting a recently discovered finite-size bound, we achieved a secure key rate of about 10 Mb/s at low losses (5 dB), and around 6.5~kb/s in the high-loss (38.5 dB), low block length ($N=10^4$) regime. The latter results demonstrate the system's suitability for highly lossy and time-constrained scenarios such as QKD from low Earth orbit satellites.

Figures (8)

• Figure 1: (A) The source is composed by a gain-switched DFB laser generating optical pulses that go through an isolator (ISO), and a polarizer (POL) that projects the light into the $\ket{D}$ polarization. The polarized pulses are injected into a free-space-coupled beam-splitter (BS): the beam gets reflected into an iPOGNAC modulator, consisting of a polarizing beamsplitter (PBS) and a phase-modulator ($\phi$-Mod), and travels back to be transmitted. The output states are once again injected to an isolator and a $\ket{D}$-aligned polarizer, effectively attenuating the optical pulses by a factor of $\cos^2(\Delta\phi)$, which are then used as input states to a second iPOGNAC modulator used for polarization encoding. The transmission port of the free-space-coupled BS for this modulation stage is coupled to a single-mode fiber (SMF), including a variable optical attenuator (VOA) to control the mean photon number of the source, and a dense wavelength-division multiplexer (DWDM) to spectrally filter the quantum signal. (B) The receiver consists of a DWDM to remove off-band noise introduced in the free-space channel, followed by a VOA to control channel losses, a fiber BS, two polarization-controllers (PC), two PBSs and four detectors, one for each polarization state (namely L, R, D, A). (C) Aerial view of the 620 m free-space channel between DFA and DEI in Padova. The source and receiver are located in the same laboratory, but we differentiate the Tx components from the Rx ones as LUX-Tx and LUX-Rx. Data from Google Earth [©2025 Google]. (D) Schematic of the Tx terminal at DFA, connected to LUX-Tx through a deployed fiber network. It is also shown the render of the 3D model of the enclosure hosting the optical terminal. (E) Schematic of the optical components at Rx terminal. It is also shown the render of the 3D model of the receiving telescope. TTM: tip-tilt mirror; CAM: camera; M: 2-inch mirror; FSM: fast-steering mirror; DM: dichroic mirror; BF: bulk filter; SMF: single-mode fiber; PSD: position-sensitive-detector; PM: power meter.
• Figure 2: Measured $\text{QBER}_B^k$ over time for a 1024-symbol pseudo-random sequence, where $B \in \{Y,X\}$ denotes the key and check bases, respectively, and $k \in \{\mu,\nu\}$ the signal and decoy states.
• Figure 3: Estimated at $5$ dB channel losses for the three-state one-decoy BB84 protocol at $1.5$ GHz repetition rate (green line, upper axis). In the lower axis, the measured for both key and check bases (in red and blue respectively).
• Figure 4: Estimated as a function of channel attenuation. The dashed line indicates the asymptotic limit, while the red and blue points correspond to finite-size analyses using the Hoeffding and Chernoff bounds, respectively. Inset: Difference in the estimated key rate between the Chernoff and Hoeffding bounds.
• Figure 5: Centroids displacement on the PSD when (A) FSM-OFF and (B) FSM-ON. Resulting system coupling efficiency with (C) no correction and (D) correcting the effect of turbulence.