Publications

First field trial of LLM-powered AI agent for lifecycle management of autonomous driving optical networks
Xiaomin Liu, Qizhi Qiu, Yihao Zhang, Yuming Cheng, Lilin Yi, Weisheng Hu, Qunbi Zhuge*

• Accepted by Optical Fiber Communication Conference (OFC) 2025
• Built the world’s first field trial demonstration of the LLM-powered AI agent for the network lifecycle.
• Supported diverse network events, including wavelength add/drop, failure management, and power optimization.
• Showed the potential of LLMs to control and manage autonomous optical networks.

Auto-DTWave: Digital Twin-Aided Autonomous Optical Network Operation with Continuous Wavelength Loading
Xiaomin Liu, Qizhi Qiu, Yihao Zhang, Meng Cai, Yichen Liu, Lilin Yi, Weisheng Hu, Qunbi Zhuge*

• Accepted by Optical Fiber Communication Conference (OFC) 2024
• Developed joint online digital twin (DT) construction and amplifier configuration with continuous wavelength loading in a commercial testbed. The DT achieves an RMSE of 0.37 dB, assisting near-optimal amplifier configuration with less than 0.1 dB average Q-factor deviation.

Digital Twin Modeling and controlling of Optical Power Evolution Enabling Autonomous-driving Optical Networks: A Bayesian Approach
Xiaomin Liu, Yihao Zhang, Yuli Chen, Yichen Liu, Meng Cai, Qizhi Qiu, Mengfan Fu, Lilin Yi, Weisheng Hu, Qunbi Zhuge*

• Accepted by Advanced Photonics (IF = 20.6).
• Proposed a Bayesian inference framework (BIF) to construct the digital twin of OAs and control OPE in a data-efficient manner. Only informative data are collected to balance exploration and exploitation, enabling efficient autonomous-driving optical networks (ADONs).

SMOF: Simultaneous Modeling and Optimization Framework for Raman Amplifiers in C+L-band Optical Networks
Xiaomin Liu, Yihao Zhang, Meng Cai, Yichen Liu, Lilin Yi, Weisheng Hu, Qunbi Zhuge*

• Accepted by Journal of Lightwave Technology (JLT), extended from ECOC 2023.
• Proposed a simultaneous modeling and optimization framework (SMOF) to model and optimize the gain profile of an RA in a data-efficient manner.

Building A Digital Twin for Intelligent Optical Networks [Invited Tutorial]
Qunbi Zhuge, Xiaomin Liu, Yihao Zhang, Meng Cai, Yichen Liu, Qizhi Qiu, Xueying Zhong, Jiaping Wu, Ruoxuan Gao, Lilin Yi, Weisheng Hu

• Accepted by Journal of Optical Communication and Networking (JOCN).
• Introduced and discussed three key technologies, including modeling, telemetry, and self-learning, to build a DT for optical networks.

Fusing Physics to Fiber Nonlinearity Model for Optical Networks Based on Physics-Guided Neural Networks
Xiaomin Liu, Yunyun Fan, Yihao Zhang, Meng Cai, Lei Liu, Lilin Yi, Weisheng Hu, Qunbi Zhuge*

• Accepted by Journal of Lightwave Technology (JLT), extended from ACP 2021 as the Best student paper.
• Proposed a fiber model based on physics-guided neural networks (PGNN). The PGNN-based model is compared with traditional neural networks in an experimental link to further illustrate its superior accuracy and generalizability.

A Meta-Learning-Assisted Training Framework for Physical Layer Modeling in Optical Networks
Xiaomin Liu, Huazhi Lun, Lei Liu, Yihao Zhang, Yichen Liu, Lilin Yi, Weisheng Hu, Qunbi Zhuge*

• Accepted by Journal of Lightwave Technology (JLT), extended from ECOC 2020.
• Proposed a meta-learning-assisted training framework for machine-learning-based physical layer models. This framework improves model robustness to agnostic uncertain parameters during offline training and enables the model to efficiently adapt to the real system with fewer data.

A Data-Fusion-Assisted Telemetry Layer for Autonomous Optical Networks [Invited Paper]
Xiaomin Liu, Huazhi Lun, Ruoxuan Gao, Meng Cai, Lilin Yi, Weisheng Hu, Qunbi Zhuge*

• Accepted by Journal of Lightwave Technology (JLT).
• Proposed a data-fusion-assisted telemetry layer between the physical layer and control layer. The data fusion methodologies are elaborated at three different levels: source level, space level, and model level.