基于LSTM的5G-TSN时钟同步误差预测

刘青俊; 王然; 黄震宁; 郝洁; 滕艺丹; 吴强

doi:10.11959/j.issn.2096-109x.2026004

您当前的位置：

首页 >

文章列表页 >

基于LSTM的5G-TSN时钟同步误差预测

学术论文 | 更新时间：2026-05-26

- 基于LSTM的5G-TSN时钟同步误差预测
- 5G-TSN clock synchronization error prediction based on long short-term memory
- 网络与信息安全学报 2026年12卷第1期页码：40-53
- 作者机构：
  
  1.南京航空航天大学计算机科学与技术学院，江苏南京 211106
  2.中国移动通信集团有限公司研究院，北京 100032
  3.深圳职业技术大学人工智能学院，广东深圳 518055
  4.浙江大学计算机科学与技术学院，浙江杭州 310027
- 作者简介：
  
  刘青俊（2001- ），男，山西大同人，南京航空航天大学硕士生，主要研究方向为确定性网络、工业互联网。
  王然（1989- ），男，黑龙江哈尔滨人，南京航空航天大学副教授、博士生导师，主要研究方向为云计算、工业互联网、人工智能应用等。
  黄震宁（1988- ），男，安徽枞阳人，中国移动通信集团有限公司研究院高级工程师，主要研究方向为5G/6G网络演进、工业互联网、能力开放、智能化、确定性网络等。
  郝洁（1988- ），女，安徽亳州人，南京航空航天大学副教授、硕士生导师，主要研究方向为无线传感、无线通信等。
  滕艺丹（1987- ），女，吉林吉林人，深圳职业技术大学讲师，主要研究方向为图像处理、增强现实与虚拟现实技术应用。
  吴强（1973- ），男，新疆伊宁人，浙江大学教授、博士生导师，主要研究方向为未来网络、工业互联网、空天地一体化网络等。
- 基金信息：
  
  国家自然科学基金(62171218);江苏省前沿技术研发项目(BF2024067)
- DOI：10.11959/j.issn.2096-109x.2026004
  中图分类号： TP393
- 收稿：2025-06-23，
  
  修回：2025-12-11，
  
  纸质出版：2026-02-25
- 稿件说明：
移动端阅览
刘青俊, 王然, 黄震宁, 等. 基于LSTM的5G-TSN时钟同步误差预测[J]. 网络与信息安全学报, 2026, 12(1): 40-53.

Liu Qingjun, Wang Ran, Huang Zhenning, et al. 5G-TSN clock synchronization error prediction based on long short-term memory[J]. Chinese Journal of Network and Information Security, 2026, 12(1): 40-53.
刘青俊, 王然, 黄震宁, 等. 基于LSTM的5G-TSN时钟同步误差预测[J]. 网络与信息安全学报, 2026, 12(1): 40-53. DOI： 10.11959/j.issn.2096-109x.2026004.

Liu Qingjun, Wang Ran, Huang Zhenning, et al. 5G-TSN clock synchronization error prediction based on long short-term memory[J]. Chinese Journal of Network and Information Security, 2026, 12(1): 40-53. DOI： 10.11959/j.issn.2096-109x.2026004.

摘要

5G与时间敏感网络（time-sensitive networking

TSN）融合架构被视为工业4.0的核心使能技术，5G-TSN端到端确定性传输依赖高精度的时钟同步。然而，5G与TSN之间存在时钟异构性，特别是基站与TSN时间转换设备间的时钟频率偏移传递性，使得系统级同步误差呈现时变非线性特征，现有研究缺乏有效的同步误差动态监测与预测机制。针对此问题，建立了5G-TSN时钟同步模型，并提出基于长短期记忆（long short-term memory

LSTM）的时钟同步误差预测机制。LSTM具备建模长期时序依赖的能力，在结构上适应时钟误差的累计性与动态变化特征，能够有效预测趋势并识别潜在的同步异常，为同步补偿提供前馈调控支撑，提升同步可靠性。仿真实验结果表明，在典型工业场景仿真数据集上，该方法的决定系数

达0.951，在均方误差等评价指标上优于反向传播神经网络、递归神经网络、门控循环单元和基于自注意力机制的Transformer模型，为构建高精度、可调控的亚微秒级5G-TSN融合同步系统提供了理论支撑。

Abstract

The integrated architecture of the fifth-generation mobile communication technology (5G) and time-sensitive networking (TSN) is widely regarded as a core enabling technology for Industry 4.0. End-to-end deterministic transmission in 5G-TSN relies on high-precision clock synchronization. However

due to the clock heterogeneity between 5G and TSN

particularly the transferability of clock frequency offsets between base stations and TSN time conversion devices

system-level synchronization errors exhibit time-varying nonlinear characteristics. Current research lacks effective m

echanisms for dynamic monitoring and prediction of synchronization errors. To address this issue

a 5G-TSN clock synchronization model was established

and a clock synchronization error prediction mechanism based on long short-term memory (LSTM) was proposed. LSTM's ability to model long-term temporal dependencies aligns well with the cumulative and dynamic nature of clock errors

enabling it to predict trends and identify potential synchronization anomalies effectively. This provides feed forward control support for synchronization compensation and enhances reliability.Simulation experiment results show that on a typical industrial scene simulation dataset

the determination coefficient

of this method reaches 0.951

and it outperforms backpropagation neural networks

recurrent neural networks

gated recurrent units

and Transformer models based on self-attention mechanisms in terms of evaluation metrics such as mean squared error. This work provides theoretical support for constructing high-precision

adjustable sub-microsecond 5G-TSN integrated synchronization systems.

关键词

Keywords

references

Rikalovic A , Suzic N , Bajic B , et al . Industry 4.0 implementation challenges and opportunities: a technological perspective [J ] . IEEE Systems Journal , 2022 , 16 ( 2 ): 2797 - 2810 .

5G Alliance for Connected Industrials and Automation . 5G for connected industries and automation(second edition) [EB ] . 2025 .

Lo Bello L , Steiner W . A perspective on IEEE time-sensitive networking for industrial communication and automation systems [J ] . Proceedings of the IEEE , 2019 , 107 ( 6 ): 1094 - 1120 .

Wu H . Ten reflections on 5G [J ] . ZTE communications , 2020 , 18 ( 1 ): 1 - 4 .

3GPP. System architecture for the 5G system (5GS): 3GPP TS 23.501 version 16.10.0 [S ] . 2020 .

Acia G . Integration of 5G with time-sensitive networking for industrial communications [R ] . 2021 .

3GPP. System architecture for the 5G system: 3GPP TS 23.501 V15.3.0 [S ] . 2018 .

Jiang X , Luvisotto M , Pang Z , et al . Reliable minimum cycle time of 5G NR based on data-driven channel characterization [J ] . IEEE Transactions on Industrial Informatics , 2021 , 17 ( 11 ): 7401 - 7411 .

Au E . A short update on 3GPP release 16 and release 17 standards [J ] . IEEE Vehicular Technology Magazine , 2020 , 15 ( 2 ): 160 - 160 .

朱瑾瑜 , 张恒升 , 陈洁 . TSN 与 5G 融合部署的需求和网络架构演进 [J ] . 中兴通讯技术 , 2021 , 27 ( 6 ): 47 - 52 .

Zhu J Y , Zhang H S , Chen J , et al . Requirements and network architecture evolution of integrated TSN and 5G deployments [J ] . ZTE Communications , 2021 , 27 ( 6 ): 47 - 52 .

Muslim A B , TöNjes R , Bauschert T . Synchronizing TSN devices via 802.1 AS over 5G networks [J ] . Electronics , 2024 , 13 ( 4 ): 768 .

李晓辉 , 王先文 , 樊韬 , 等 . 5G-TSN系统下的高精度时间同步 [J ] . 系统工程与电子技术 , 2023 , 45 ( 2 ): 559 - 565 .

Li X H , Wang X W , Fan T , et al . High-precision time synchronization in 5G-TSN systems [J ] . Systems Engineering Electronics , 2023 , 45 ( 2 ): 559 - 565 .

Song J , Kubomi M , Zhao J , et al . Time synchronization performance analysis considering the frequency offset inside 5G-TSN network [C ] // Proceedings of the 2021 17th International Symposium on Wireless Communication Systems . Piscataway : IEEE Press , 2021 : 1 - 6 .

Chai Z C , Liu W , Li M , et al . Cross domain clock synchronization based on data packet relay in 5G-TSN integrated network [C ] // Proceedings of the 2021 IEEE 4th International Conference on Electronics and Communication Engineering (ICECE) . Piscataway : IEEE Press , 2022 : 141 - 145 .

Striffler T , Schotten H D . The 5G transparent clock: synchronization errors in integrated 5G-TSN industrial networks [C ] // Proceedings of the 2021 IEEE 19th International Conference on Industrial Informatics (INDIN) . Piscataway : IEEE Press , 2021 : 1 - 6 .

Wei X H , Wei X Z , Luo Z Q , et al . An enhanced IEEE1588 clock synchronization for link delays based on a system-on-chip platform [J ] . International Journal of Electronics and Telecommunications , 2021 : 289 - 294 .

朱海龙 , 严园园 , 张华宇 . 基于驻留时间补偿的5G-TSN系统高精度时间同步方法 [J ] . 北京邮电大学学报 , 2023 , 46 ( 6 ): 89 - 94 .

Zhu H L , Yan Y Y , Zhang H Y . et al . High-precision time synchronization method for 5G-TSN systems based on dwell time compensation [J ] . Journal of Beijing University of Posts and Telecommunications , 2023 , 46 ( 6 ): 89 - 94 .

Wang J Q , Du Y , Wang J . LSTM based long-term energy consumption prediction with periodicity [J ] . Energy , 2020 , 197 : 117197 .

Pano-Azucena A D , Tlelo-Cuautle E , Tan S X D . Prediction of chaotic time series by using ANNs, ANFIS and SVMs [C ] // Proceedings of the 2018 7th International Conference on Modern Circuits and Systems Technologies (MOCAST) . Piscataway : IEEE Press , 2018 : 1 - 4 .

Aishwarya D C , Babu C N . Prediction of time series data using GA-BPNN based hybrid ANN model [C ] // Proceedings of the 2017 IEEE 7th International Advance Computing Conference (IACC) . Piscataway : IEEE Press , 2017 : 848 - 853 .

Hochreiter S , Schmidhuber J . Long short-term memory [J ] . Neural Computation , 1997 , 9 ( 8 ): 1735 - 1780 .

Vaswani A , Shazeer N , Parmar N , et al . Attention is all you need [C ] // Advances in Neural Information Processing Systems . 2017 : 5998 - 6008 .

浏览量

207

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于面部特征点运动的活体识别方法

VeriFU：可验证的联邦遗忘学习

基于告警增量分析的攻击场景实时重构方法

基于流量交互图分析的Web服务威胁检测方法

SI-CP-ABE：一类基于同源的密文策略属性基加密方案