Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

Frontiers in Optics

Physics-informed neural networks for classical and quantum communication with multimode fibers

Qian Zhang, Jiali Sun, Stefan Rothe, Jürgen W. Czarske

Published Jan 1, 2024• 1 citation
Publisher

Abstract

Physics-informed neural networks provide mode decomposition based on intensity-only images in high-dimensional quantum communication using multimode fibers.
DOI
10.1364/fio.2024.fm4b.3

Related papers

High-dimensional encryption in optical fibers using spatial modes of light and machine learning

M. L. Lollie, Fatemeh Mostafavi, Narayan Bhusal +5 more

Abstract The ability to engineer the spatial wavefunction of photons has enabled a variety of quantum protocols for communication, sensing, and information processing. These protocols exploit the high dimensionality of structured light enabling the encoding of multiple bits of information in a single photon, the measurement of small physical parameters, and the achievement of unprecedented levels of security in schemes for cryptography. Unfortunately, the potential of structured light has been restrained to free-space platforms in which the spatial profile of photons is preserved. Here, we make an important step forward to using structured light for fiber optical communication. We introduce a classical encryption protocol in which the propagation of high-dimensional spatial modes in multimode fibers is used as a natural mechanism for encryption. This provides a secure communication channel for data transmission. The information encoded in spatial modes is retrieved using artificial neural networks, which are trained from the intensity distributions of experimentally detected spatial modes. Our on-fiber communication platform allows us to use single spatial modes for information encoding as well as the high-dimensional superposition modes for bit-by-bit and byte-by-byte encoding respectively. This protocol enables one to recover messages and images with almost perfect accuracy. Our classical smart protocol for high-dimensional encryption in optical fibers provides a platform that can be adapted to address increased per-photon information capacity at the quantum level, while maintaining the fidelity of information transfer. This is key for quantum technologies relying on structured fields of light, particularly those that are challenged by free-space propagation.

arXiv:2108.06420•Machine Learning Science and Technology•Jul 6, 2022•16 citations

SOLVING QUANTUM EIGENVALUE PROBLEMS USING PHYSICS- INFORMED NEURAL NETWORKS

IDRIS ODUOLA

•Jan 1, 2024•0 citations

Multitask parallel optical communication: Driven by high-dimensional orbital angular momentum arrays in multimode fibers

Jingjing Wang, Xianjing Li, Jiawei Wu +5 more

As the demand for data capacity continues to grow, optical fiber communication systems are progressively approaching their theoretical limits, thereby motivating the exploration of high-dimensional (HD) degrees of freedom such as orbital angular momentum (OAM). However, strong intermodal coupling in multimode fibers (MMFs) poses substantial challenges for the transmission and recovery of HD optical information. Here, we propose and experimentally demonstrate an HD OAM-array-based multichannel multitask optical communication framework via a single MMF. By encoding different information tasks into HD OAM superposition states across multiple wave-vector (k) channels, the proposed scheme enables the simultaneous transmission of heterogeneous optical information within a unified fiber platform. At the receiver, a residual neural network constrained by a multi-channel quantum-state fidelity loss function directly reconstructs the transmitted signals from single-frame speckle intensity measurements, without requiring phase retrieval or channel inversion. Experiments demonstrate the synchronous communication of multichannel classical image signals and HD mutually unbiased basis states through the same MMF, while maintaining high reconstruction fidelity and low interchannel crosstalk under strong modal mixing conditions. This approach provides a scalable route toward high-capacity hybrid classical–quantum optical communication in fiber systems.

Applied Physics Letters•Oct 5, 2026•0 citations

Inverse physics-informed neural networks for noise detection in open quantum systems

Gubio Gomes de Lima, Iann Cunha, L. K. Castelano

arXiv:2507.12552•Physics Letters A•Mar 7, 2026•3 citations

Experimental demonstration of deep-learning-enabled adaptive optics

Hao-Bin Fu, Zu-Yang Wan, Yu-Huai Li +9 more

Satellite-based quantum communication is a promising approach for establishing global-scale quantum networks. In free-space quantum channels, single-mode-fiber coupling plays a crucial role in increasing the signal-to-noise ratio of daylight quantum key distribution (QKD) and ensuring compatibility with standard fiber-based QKD protocols. However, consistently achieving high efficiency and stable single-mode-fiber coupling under strong atmospheric turbulence remains an ongoing experimental challenge. In this study, we experimentally demonstrate an adaptive method based on convolutional neural networks capable of directly estimating phase information from a single defocused image. We developed a convolutional neural network to establish the relationship between intensity distribution and phase information of turbulent distortions. We demonstrate the real-time performance of our deep-learning adaptive method in increasing single-mode-fiber coupling efficiency across various turbulence scales and quantify turbulence frequencies. Notably, the method proved highly effective in strong-turbulence scenarios, with frequencies reaching up to 200 Hz, leading to a significant increase in single-mode-fiber coupling efficiency. We demonstrate the corrective capability of our adaptive method for strong turbulence, enabled by the generalization of the convolutional neural network. Our results offer an efficient solution for daytime free-space QKD applications.

Physical Review Applied•Sep 19, 2024•11 citations

Quantum Optical Neural Network with Multimode Cavity QED

Yudan Guo, Brendan Marsh, Ronen M. Kroeze +3 more

Bulletin of the American Physical Society•Mar 5, 2020•0 citations
Abstract The ability to engineer the spatial wavefunction of photons has enabled a variety of quantum protocols for communication, sensing, and information processing. These protocols exploit the high dimensionality of structured light enabling the encoding of multiple bits of information in a single photon, the measurement of small physical parameters, and the achievement of unprecedented levels of security in schemes for cryptography. Unfortunately, the potential of structured light has been restrained to free-space platforms in which the spatial profile of photons is preserved. Here, we make an important step forward to using structured light for fiber optical communication. We introduce a classical encryption protocol in which the propagation of high-dimensional spatial modes in multimode fibers is used as a natural mechanism for encryption. This provides a secure communication channel for data transmission. The information encoded in spatial modes is retrieved using artificial neural networks, which are trained from the intensity distributions of experimentally detected spatial modes. Our on-fiber communication platform allows us to use single spatial modes for information encoding as well as the high-dimensional superposition modes for bit-by-bit and byte-by-byte encoding respectively. This protocol enables one to recover messages and images with almost perfect accuracy. Our classical smart protocol for high-dimensional encryption in optical fibers provides a platform that can be adapted to address increased per-photon information capacity at the quantum level, while maintaining the fidelity of information transfer. This is key for quantum technologies relying on structured fields of light, particularly those that are challenged by free-space propagation.
As the demand for data capacity continues to grow, optical fiber communication systems are progressively approaching their theoretical limits, thereby motivating the exploration of high-dimensional (HD) degrees of freedom such as orbital angular momentum (OAM). However, strong intermodal coupling in multimode fibers (MMFs) poses substantial challenges for the transmission and recovery of HD optical information. Here, we propose and experimentally demonstrate an HD OAM-array-based multichannel multitask optical communication framework via a single MMF. By encoding different information tasks into HD OAM superposition states across multiple wave-vector (k) channels, the proposed scheme enables the simultaneous transmission of heterogeneous optical information within a unified fiber platform. At the receiver, a residual neural network constrained by a multi-channel quantum-state fidelity loss function directly reconstructs the transmitted signals from single-frame speckle intensity measurements, without requiring phase retrieval or channel inversion. Experiments demonstrate the synchronous communication of multichannel classical image signals and HD mutually unbiased basis states through the same MMF, while maintaining high reconstruction fidelity and low interchannel crosstalk under strong modal mixing conditions. This approach provides a scalable route toward high-capacity hybrid classical–quantum optical communication in fiber systems.
Satellite-based quantum communication is a promising approach for establishing global-scale quantum networks. In free-space quantum channels, single-mode-fiber coupling plays a crucial role in increasing the signal-to-noise ratio of daylight quantum key distribution (QKD) and ensuring compatibility with standard fiber-based QKD protocols. However, consistently achieving high efficiency and stable single-mode-fiber coupling under strong atmospheric turbulence remains an ongoing experimental challenge. In this study, we experimentally demonstrate an adaptive method based on convolutional neural networks capable of directly estimating phase information from a single defocused image. We developed a convolutional neural network to establish the relationship between intensity distribution and phase information of turbulent distortions. We demonstrate the real-time performance of our deep-learning adaptive method in increasing single-mode-fiber coupling efficiency across various turbulence scales and quantify turbulence frequencies. Notably, the method proved highly effective in strong-turbulence scenarios, with frequencies reaching up to 200 Hz, leading to a significant increase in single-mode-fiber coupling efficiency. We demonstrate the corrective capability of our adaptive method for strong turbulence, enabled by the generalization of the convolutional neural network. Our results offer an efficient solution for daytime free-space QKD applications.