Distribution of non-Gaussian states in a deployed telecommunication fiber channel

TL;DR

The paper addresses the challenge of distributing non-Gaussian continuous-variable quantum states over real-world telecom fiber networks. It generates photon-subtracted squeezed states (Schrödinger kitten states) at 1550 nm and transmits them across ~300 m of deployed campus fiber, validating non-Gaussianity with quantum state tomography. Loss-corrected measurements show persistent negativity in the Wigner function across the link, e.g., $W(0,0)$ reaching $-0.164(4)$ locally and $-0.028(4)$ after transmission, with initial fidelity to an ideal cat of $66\%$ (α = 0.91) decreasing to ~52–53% after transmission. These results demonstrate the practical viability of distributing non-Gaussian bosonic codes on real networks and outline a path toward networked quantum information processing with potential Bell-inequality violations and quantum steering, while highlighting the need for quantum error-correction integration for long-distance quantum networking.

Abstract

Optical non-Gaussian states hold great promise as a pivotal resource for advanced optical quantum information processing and fault-tolerant long-distance quantum communication. Establishing their faithful transmission in a real-world communication channel, therefore, marks an important milestone. In this study, we experimentally demonstrate the distribution of such non-Gaussian states in a functioning telecommunication channel that connects separate buildings within the DTU campus premises. We send photon-subtracted squeezed states, exhibiting pronounced Wigner negativity, through 300 m of deployed optical fibers to a distant building. Using quantum homodyne tomography, we fully characterize the states upon arrival. Our results show the survival of the Wigner function negativity after transmission when correcting for detection losses, indicating that the established link can potentially facilitate the violation of Bell's inequality and enable quantum steering. This achievement not only validates the practical feasibility of distributing non-Gaussian states in real-world settings, but also provides an exciting impetus towards realizing fully coherent quantum networks for high-dimensional, continuous-variable quantum information processing.

Figures (8)

• Figure 1: The Schrödinger kitten states are generated in a basement lab of DTU's building 307 (A) and distributed through $\sim$370m of fiber installed in DTU's tunnel system -- as roughly indicated by the pink path -- to a ground-level technical room in building 340 (B) via a switchboard in building 341. The optical link is made up of five SC-connectorized single-mode fiber patches. The central three patches of about 300m length (indicated by the hashed block in the diagram) are part of the already deployed campus network, while the first and last patches were set up for this demonstration. Due to the various connections at the intermediate stations, the optical states experience a loss of around 22% during transmission. Map data: Google, Landsat / Copernicus.
• Figure 2: Experimental setup. SHG: second harmonic generation, OPO: optical parametric oscillator, BPF: bandpass filter, SPD: single photon detection, LO: local oscillator, E$\rightarrow$O: electrical to optical converter, O$\rightarrow$E: optical to electrical converter, (DE)MUX: optical (de)multiplexer, QST: quantum state tomography, HD: homodyne detection, DSO: digital storage oscilloscope. Please see main text for details.
• Figure 3: (a) Raw homodyne traces as a function of time relative to the trigger signal. The elevated noise region preceding the trigger corresponds to the temporal mode where a single photon has been subtracted from the squeezed vacuum state, leading to the generation of the non-Gaussian kitten state. (b) Histogram of quadrature measurements for the anti-squeezed p-quadrature using the optimal temporal mode function $f(t)$. The solid orange histogram corresponds to the photon-subtracted squeezed state, while the dotted histogram shows the reference distribution from the non-subtracted squeezed vacuum. Both distributions are normalized to the shot-noise level, which itself is indicated by the vacuum state histogram in solid blue.
• Figure 4: Reconstructed Wigner functions of the generated states. Wigner functions of the squeezed vacuum state (left), the photon-subtracted squeezed state (middle), and the loss-corrected photon-subtracted state (right) measured locally at the source in building A. The initial squeezed vacuum state shows Gaussian statistics, while the photon-subtracted state exhibits a pronounced dip in the Wigner function, $W(0,0) = -0.108(5)$, a hallmark of its non-Gaussian nature. Correcting for the 12% detection loss further enhances the negativity to $W(0,0) = -0.164(4)$.
• Figure 5: Reconstructed Wigner functions of the states transmitted across the campus fiber network. Wigner functions of the squeezed vacuum state (left), the photon-subtracted squeezed state (middle), and the loss-corrected photon-subtracted state (right) measured at the receiver in building B. When corrected for the 12% inefficiency attributed purely to the homodyne detection at the receiving station, the transmitted state maintains a bit of negativity, with $W(0,0) = -0.028(4)$. Without this correction (middle), we have $W(0,0) = 0.006(5)$.