光通信中非线性均衡方法与泛化性研究进展

张天倩; 何擎宇; 罗鸣; 李重阳; 刘守印

doi:10.13756/j.gtxyj.2026.250313

您当前的位置：

首页 >

文章列表页 >

光通信中非线性均衡方法与泛化性研究进展

光通信系统与网络技术 | 更新时间：2026-10-21

- 光通信中非线性均衡方法与泛化性研究进展
- Research Progress on Nonlinear Equalization Methods and Generalization in Optical Communication
- 光通信研究 2026年第5期页码：64-74
- 作者机构：
  
  1.中国信息通信科技集团有限公司　光通信技术和网络全国重点实验室，武汉　430074
  2.华中师范大学　物理科学与技术学院，武汉　4300743
  3.鹏城实验室，广东　深圳　518055
- 作者简介：
  
  张天倩（2000-），女，河南南阳人。博士，主要研究方向为光通信传输算法。
  罗鸣，教授级高级工程师。E-mail：mluo@cict.com
- 基金信息：
  
  湖北省技术创新计划资助项目(2024BAA001);湖北省中央引导地方科技发展专项资助项目(2024CSA077)
- DOI：10.13756/j.gtxyj.2026.250313
  中图分类号： TN929
- 收稿：2025-09-06，
  
  修回：2025-10-10，
  
  纸质出版：2026-10-10
- 稿件说明：
移动端阅览
张天倩，何擎宇，罗鸣，等. 光通信中非线性均衡方法与泛化性研究进展[J]. 光通信研究，2026(5): 64-74.

Zhang T Q, He Q Y, Luo M, et al. Research Progress on Nonlinear Equalization Methods and Generalization in Optical Communication[J]. Study on Optical Communications, 2026(5): 64-74.
张天倩，何擎宇，罗鸣，等. 光通信中非线性均衡方法与泛化性研究进展[J]. 光通信研究，2026(5): 64-74. DOI： 10.13756/j.gtxyj.2026.250313.

Zhang T Q, He Q Y, Luo M, et al. Research Progress on Nonlinear Equalization Methods and Generalization in Optical Communication[J]. Study on Optical Communications, 2026(5): 64-74. DOI： 10.13756/j.gtxyj.2026.250313.

摘要

目的

随着相干光通信系统向更高容量、更长距离和更高阶调制演进，光纤非线性效应已成为制约系统性能的主要瓶颈，非线性均衡（NLE）成为提升系统性能的关键研究方向。近年来，机器学习方法在光通信中的应用不断深化，在复杂非线性建模与补偿方面展现出强大优势，但这些方法在计算复杂度和泛化性方面仍面临挑战。文章旨在系统描述传统非线性补偿方法与基于机器学习的非线性均衡方法，并分析提升泛化性与降低复杂度的策略。

方法

首先介绍传统数字信号处理（DSP）方法，包括数字反向传播（DBP）和基于扰动理论的补偿（PBC）的基本原理与复杂度。接着介绍多层感知器（MLP）、卷积神经网络（CNN）、双向长短期记忆网络（Bi-LSTM）、Transformer、生成对抗网络（GAN）以及端到端深度学习（E2E-DL）等神经网络均衡器的研究进展，并给出复杂度计算公式。最后讨论剪枝/权重聚类/量化和知识蒸馏，迁移学习/元学习以及半监督学习/自监督学习等降低复杂度与提升泛化性的多种技术。

结论

目前基于机器学习的非线性均衡方法在性能上已逐渐超越传统DSP方案，但其复杂度与能效问题仍是制约实际部署的关键。未来研究需要在性能、复杂度与泛化性之间寻求平衡，以推动基于机器学习的非线性均衡从实验验证走向实际应用。

Abstract

Objective

With the continuous evolution of coherent optical communication systems toward higher capacity

longer transmission distances

and higher-order modulation formats

fiber nonlinear effects have become a major bottleneck limiting the system performance. Nonlinear Equalization (NLE) has emerged as a key technique to enhance overall transmission performance. In recent years

machine learning approaches have been increasingly applied to optical fiber communications

demonstrating unique advantages in modeling and compensating complex nonlinearities. However

these methods still face challenges in computational complexity and generalization. This paper aims to provide a systematic review of both traditional nonlinear compensation techniques and machine learning-based NLE methods

with a particular focus on strategies for enhancing generalization and reducing complexity.

Methods

We first review traditional Digital Signal Processing (DSP) approaches

including the basic principles and computational complexity of Digital Back Proagation (DBP) and Perturbation-Based Compensation (PBC). We then summarize recent advances in neural network-based equalizers

such as Multilayer Perceptrons (MLP)

Convolutional Neural Networks (CNN)

bidirectional long short-term memory networks (Bi-LSTM)

Transformers

Generative Adversarial Networks (GAN)

and End-to-end Deep Learning (E2E-DL)

along with their corresponding complexity formulations. Finally

we highlight various techniques for complexity reduction and generalization enhancement

including pruning

weight clustering

quantization

knowledge distillation

transfer learning

meta-learning

semi-supervised learning

and self-supervised learning.

Conclusion

Overall

machine learning based NLE techniques have gradually outperformed traditional DSP solutions in terms of compensation performance. Nevertheless

challenges in computational complexity and energy efficiency remain major obstacles to real-world deployment. Future research should aim to balance performance

complexity

and generalization

thereby facilitating the transition of machine learning-based NLE algorithms from proof-of-concept studies to practical applications.

关键词

Keywords

references

Mitra P P , Stark J B . Nonlinear Limits to the Information Capacity of Optical Fiber Communications [J ] . Nature , 2001 , 411 ( 6841 ): 1027 - 1030 .

Ip E . Nonlinear Compensation Using Backpropagation for Polarization-Multiplexed Transmission [J ] . Journal of Lightwave Technology , 2010 , 28 ( 6 ): 939 - 951 .

Li C , Zhang F , Zhu Y , et al . Fiber Nonlinearity Mitigation in Single Carrier 400 G and 800 G Nyquist-WDM Systems [J ] . Journal of Lightwave Technology , 2018 , 36 ( 17 ): 3707 - 3715 .

Spinnler B . Equalizer Design and Complexity for Digital Coherent Receivers [J ] . IEEE Journal of Selected Topics in Quantum Electronics , 2010 , 16 ( 5 ): 1180 - 1192 .

Sidelnikov O , Redyuk A , Sygletos S . Equalization Performance and Complexity Analysis of Dynamic Deep Neural Networks in Long Haul Transmission Systems [J ] . Optics Express , 2018 , 26 ( 25 ): 32765 .

Li L , Tao Z , Dou L , et al . Implementation Efficient Nonlinear Equalizer based on Correlated Digital Backpropagation [C ] // Optical Fiber Communication Conference/National Fiber Optic Engineers Conference 2011 . Los Angeles, California : OSA , 2011 : OWW3 .

Yan W , Tao Z , Dou L , et al . Low Complexity Digital Perturbation Back-Propagation [C ] // 37th European Conference and Expositionon Optical Communications . Geneva : OSA , 2011 : Tu.3.A.2 .

Bütler R M , Häger C , Pfister H D , et al . Model-based Machine Learning for Joint Digital Backpropagation and PMD Compensation [J ] . Journal of Lightwave Technology , 2021 , 39 ( 4 ): 949 - 959 .

陈志轩 , 蔡炬 , 张洪波 , 等 . 基于LDBP的色散和光纤非线性损伤补偿 [J ] . 光通信研究 , 2024 ( 4 ): 230047 .

Chen Z X , Cai J , Zhang H B , et al . Chromatic Dispersion and Fiber Nonlinear Impairment Compensation based on LDBP [J ] . Study on Optical Communications , 2024 ( 4 ): 230047 .

Chi X , Bai C , Yang F , et al . Joint Intra and Inter-Channel Nonlinear Compensation Scheme based on Improved Learned Digital back Propagation for WDM Systems [J ] . Optics Express , 2024 , 32 ( 4 ): 5095 .

Mecozzi A , Clausen C B , Shtaif M . Analysis of Intra-channel Nonlinear Effects in Highly Dispersed Optical Pulse Transmission [J ] . IEEE Photonics Technology Letters , 2000 , 12 ( 4 ): 392 - 394 .

Tao Z , Dou L , Yan W , et al . Multiplier-Free Intra-channel Nonlinearity Compensating Algorithm Operating at Symbol Rate [J ] . Journal of Lightwave Technology , 2011 , 29 ( 17 ): 2570 - 2576 .

潘雨彤 , 张天泓 , 舒画 , 等 . 高速光通信系统光纤非线性均衡技术进展 [J ] . 光学学报 , 2025 , 45 ( 13 ): 1306015 .

Pan Y T , Zhang T H , Shu H , et al . Advancesin Fiber Nonlinear Compensation Technology for High-Speed Optical Communication Systems (Invited) [J ] . Acta Optica Sinica , 2025 , 45 ( 13 ): 1306015 .

Zhang S , Yaman F , Nakamura K , et al . Field and Lab Experimental Demonstration of Nonlinear Impairment Compensation Using Neural Networks [J ] . Nature Communications , 2019 , 10 : 3033 .

Li C , Wang Y , Wang J , et al . Convolutional Neural Network-Aided DP-64 QAM Coherent Optical Communication Systems [J ] . Journal of Lightwave Technology , 2022 , 40 ( 9 ): 2880 - 2889 .

范润哲 , 杨超 , 罗鸣 . 30 Tbit/s DP-16QAM传输系统中基于机器学习的非线性均衡方法 [J ] . 光通信研究 , 2024 ( 4 ): 230043 .

Fan R Z , Yang C , Luo M . Nonlinear Equalization Method based on Machine Learning in 30 Tbit/s DP-16QAM Transmission System [J ] . Study on Optical Communications , 2024 ( 4 ): 230043 .

Freire P J , Osadchuk Y , Spinnler B , et al . Performance versus Complexity Study of Neural Network Equalizers in Coherent Optical Systems [J ] . Journal of Lightwave Technology , 2021 , 39 ( 19 ): 6085 - 6096 .

Gautam N , Pendem S V , Lall B , et al . Transformer-Based Nonlinear Equalization for DP-16QAM Coherent Optical Communication Systems [J ] . IEEE Communications Letters , 2024 , 28 ( 3 ): 577 - 581 .

Osogami T , Kajino H , Sekiyama T . Bidirectional Learning for Time-Series Models with Hidden Units [C ] // Proceedings of the International Conference on Machine Learning (ICML) , Sydney, Australia : PMLR , 2017 : 2711 - 2720 .

Dai X X , Li X , Luo M , et al . LSTM Networks Enabled Nonlinear Equalization in 50-Gb/sPAM-4 Transmission Links [J ] . Applied Optics , 2019 , 58 ( 22 ): 6079 - 6084 .

Lu X , Lu C , Yu W , et al . Memory-Controlled Deep LSTM Neural Network Post-Equalizer Used in High-Speed PAM VLC System [J ] . Optics Express , 2019 , 27 ( 5 ): 7822 .

Deligiannidis S , Bogris A , Mesaritakis C , et al . Compensation of Fiber Nonlinearities in Digital Coherent Systems Leveraging Long Short-Term Memory Neural Networks [J ] . Journal of Lightwave Technology , 2020 , 38 ( 21 ): 5991 - 5999 .

Hu X , Yi X , Chen Y , et al . Inter-Channel Nonlinearity Compensation in Long-Haul Optical Fiber Communication Using Overlapped-Dense LSTM Network: Complexity Reduction and Visual Understanding [J ] . Optics Communications , 2024 , 556 : 130274 .

Xiang J , Xiang M , Lv H , et al . Low-Complexity Conditional Generative Adversarial Network (c-GAN) Based Nonlinear Equalizer for Coherent Data-Center Interconnections [J ] . Journal of Lightwave Technology , 2023 , 41 ( 18 ): 5966 - 5972 .

Karanov B , Chagnon M , Thouin F , et al . End-to-End Deep Learning of Optical Fiber Communications [J ] . Journal of Lightwave Technology , 2018 , 36 ( 20 ): 4843 - 4855 .

Karanov B , Chagnon M , Aref V , et al . Concept and Experimental Demonstration of Optical IM/DD End-to-End System Optimization Using a Generative Model [C ] // Optical Fiber Communication Conference (OFC)2020 . San Diego, California, USA : Optica Publishing Group , 2020 : Th2A.48 .

Minelli L , Forghieri F , Nespola A , et al . A Multi-Rate Approach for Nonlinear Pre-Distortion Using End-to-End Deep Learning in IM-DD Systems [J ] . Journal of Lightwave Technology , 2023 , 41 ( 2 ): 420 - 431 .

Xu Y , Huang L , Jiang W , et al . End-to-End Learning for 100G-PON based on Noise Adaptation Network [J ] . Journal of Lightwave Technology , 2024 , 42 ( 7 ): 2328 - 2337 .

Xiao Q , Ding C , Xu T , et al . Concept and Experimental Demonstration of Physics-Guided End-to-End Learning for Optical Communication Systems [J ] . Photonics Research , 2025 , 13 ( 6 ): 1469 .

Xu Z , Sun C , Ji T , et al . Feedforward and Recurrent Neural Network-Based Transfer Learning for Nonlinear Equalization in Short-Reach Optical Links [J ] . Journal of Lightwave Technology , 2021 , 39 ( 2 ): 475 - 480 .

Freire P J , Abode D , Prilepsky J E , et al . Transfer Learning for Neural Networks-Based Equalizers in Coherent Optical Systems [J ] . Journal of Lightwave Technology , 2021 , 39 ( 21 ): 6733 - 6745 .

Lennard S , Barbosa F A , Ferreira F M . Zero-Shot ML Equalization in Coherent Optical Transmission: a First Experimental Demonstration [J ] . Journal of Lightwave Technology , 2025 , 43 ( 13 ): 6194 - 6202 .

Cao Z , Wang Y , Zhang Z , et al . Meta-Learning Nonlinear Equalizer for Hyperparameters Tuning for 206.9Tbps PCS-WDM Optical Transmission [C ] // Optical Fiber Communication Conference (OFC)2025 . San Francisco, California, USA : Optica Publishing Group , 2025 : Th2A.19 .

Xiao X , Zhou Z , Dong B , et al . Meta-DSP: A Meta-Learning Approach for Data-Driven Nonlinear Compensation in High-Speed Optical Fiber Systems [J ] . Journal of Lightwave Technology , 2025 , 43 ( 16 ): 7516 - 7529 .

Zhou Q , Zhang F , Yang C . AdaNN: Adaptive Neural Network-Based Equalizer via Online Semi-Supervised Learning [J ] . Journal of Lightwave Technology , 2020 , 38 ( 16 ): 4315 - 4324 .

Tang D , Jiang Y , Wu Z , et al . Self-Supervised Learning for Neural-Network-based Perturbative Fiber Nonlinearity Compensation [J ] . IET Conference Proceedings , 2023 , 2023 ( 34 ): 218 - 221 .

Freire P J , Napoli A , Spinnler B , et al . Reducing Computational Complexity of Neural Networks in Optical Channel Equalization: From Concepts to Implementation [J ] . Journal of Lightwave Technology , 2023 , 41 ( 14 ): 4557 - 4581 .

Jiang W , Huang L , Xu Y , et al . Real-Time Deployment of Simplified Volterra Nonlinear Equalizer in High-Speed PON [J ] . IEEE Photonics Technology Letters , 2023 , 35 ( 19 ): 1067 - 1070 .

Freire P , Srivallapanondh S , Spinnler B , et al . Computational Complexity Optimization of Neural Network-Based Equalizers in Digital Signal Processing: A Comprehensive Approach [J ] . Journal of Lightwave Technology , 2024 , 42 ( 12 ): 4177 - 4201 .

Freire P J , Srivallapanondh S , Anderson M , et al . Implementing Neural Network-Based Equalizers in a Coherent Optical Transmission System Using Field-Programmable Gate Arrays [J ] . Journal of Lightwave Technology , 2023 , 41 ( 12 ): 3797 - 3815 .

Srivallapanondh S , Freire P J , Spinnler B , et al . Knowledge Distillation Applied to Optical Channel Equalization: Solving the Parallelization Problem of Recurrent Connection [C ] // Optical Fiber Communication Conference (OFC)2023 . San Diego California, USA : Optica Publishing Group , 2023 : Th1F.7 .

Zhang T Q , He Q Y , Luo M , et al . Efficient Nonlinear Equalization across S+C+L Bands Using Online Knowledge Distillation in High-Capacity DWDM Systems [J ] . Optics Express , 2025 , 33 ( 10 ): 20433 - 20448 .

浏览量

下载量

CSCD

CNKI被引量

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于多源域迁移学习的光纤非线性损伤补偿

迁移学习辅助基于机器学习的传输质量评估

基于改进EfficientNet-B0的φ-OTDR振动事件识别方法

改进YOLOv8的无人机建筑外墙缺陷检测方法

数据受限下的分布式光纤事件识别方法研究进展