基于轻量化Swin Transformer的资源自适应语义压缩策略

郑飞; 吴东颖; 李世超; 仇洪冰; 邵苏杰; 喻鹏; 丰雷; 赵继龙

doi:10.11959/j.issn.1000-436x.TXXB260128

Chinese

您当前的位置：

首页 >

文章列表页 >

基于轻量化Swin Transformer的资源自适应语义压缩策略

更新时间：2026-04-27

- 基于轻量化Swin Transformer的资源自适应语义压缩策略
- A Lightweight Swin Transformer-Based Resource-Adaptive Semantic Compression Strategy
- 通信学报 2026年
- 作者机构：
  
  1.桂林电子科技大学无线宽带通信与信号处理重点实验室，广西桂林 541004
  2.北京邮电大学网络与交换技术重点实验室，北京 100876
- 作者简介：
  
  [ "郑飞（1982- ），男，江西吉安人，博士，桂林电子科技大学副教授、硕士生导师，主要研究方向为卫星通信网、电力通信网、资源管理、移动边缘计算、确定性传输、路由规划及协议、AI大模型算法应用。" ]
  [ "吴东颖（2002- ），女，河北沧州人，桂林电子科技大学硕士生，主要研究方向为语义通信、资源管理、AI大模型算法应用。" ]
  [ "李世超（1986- ），男，甘肃兰州人，博士，桂林电子科技大学副教授、硕士生导师，主要研究方向为6G无线通信、车联网、移动边缘计算、空天地一体化网络、深度强化学习。" ]
  [ "仇洪冰（1963- ），男，江苏如皋人，博士，桂林电子科技大学教授、博士生导师，主要研究方向为宽带无线通信、通信感知一体化、通信信号处理、通信网络。" ]
  [ "邵苏杰（1985- ），男，陕西汉中人，博士，北京邮电大学副教授、博士生导师，主要研究方向为网络管理，能源互联网信息通信，边缘计算。" ]
  [ "喻鹏（1986- ），男，湖北随州人，博士，北京邮电大学副教授、博士生导师，主要研究方向为B5G/6G网络智能管控，多媒体通信管控，绿色通信，智能电网通信网运维。" ]
  [ "丰雷（1987- ），男，北京人，博士，北京邮电大学副教授、博士生导师，主要研究方向为网络智能管理、能源互联网信息通信。" ]
  [ "赵继龙（2001- ），男，山东潍坊人，桂林电子科技大学硕士生，主要研究方向为卫星通信网络资源管理、边缘计算、强化学习和人工智能。" ]
- 基金信息：
  
  国家自然科学基金资助项目(62361016);广西重点研发计划(GuikeAB25069438)
- DOI：10.11959/j.issn.1000-436x.TXXB260128
  中图分类号： TN919.8
- 收稿：2026-03-18，
  
  修回：2026-04-26，
  
  录用：2026-04-27，
- 稿件说明：
移动端阅览
郑飞, 吴东颖, 李世超, 等. 基于轻量化Swin Transformer的资源自适应语义压缩策略[J/OL]. 通信学报, 2026.

ZHENG Fei, WU Dongying, LI Shichao, et al. A Lightweight Swin Transformer-Based Resource-Adaptive Semantic Compression Strategy[J/OL]. Journal on Communications, 2026.
郑飞, 吴东颖, 李世超, 等. 基于轻量化Swin Transformer的资源自适应语义压缩策略[J/OL]. 通信学报, 2026. DOI： 10.11959/j.issn.1000-436x.TXXB260128.

ZHENG Fei, WU Dongying, LI Shichao, et al. A Lightweight Swin Transformer-Based Resource-Adaptive Semantic Compression Strategy[J/OL]. Journal on Communications, 2026. DOI： 10.11959/j.issn.1000-436x.TXXB260128.

摘要

在面向多终端的下行语义通信场景中，由于终端算力的差异性，采用统一的语义压缩比提取语义信息，会造成低算力终端难以完成语义解码，而高算力终端的算力资源未充分利用、语义数据不够精炼的问题。为解决上述问题，针对有限的终端算力资源和链路带宽，本文提出了一种基于轻量化Swin Transformer的资源自适应语义压缩策略，用门控网络和稀疏注意力机制改进了语义编码器结构，为终端定制差异化语义压缩比，以最小化系统总能耗为优化目标，构建联合算力和带宽分配模型，采用近端策略优化（Proximal Policy Optimization

PPO）算法求解优化问题。仿真结果表明，与固定语义压缩比方案相比，所提策略的系统总能耗降低了39%。

Abstract

In multi-terminal downlink semantic communication scenarios

employing a uniform semantic compression ratio caused difficulty in semantic decoding for low-computing-power terminals

as well as underutilized computing power resources and insufficiently refined semantic data for high-computing-power terminals. To address this issue under constraints of limited terminal computing power and link bandwidth

this paper proposed a lightweight Swin Transformer-based resource-adaptive semantic compression strategy. The semantic encoder integrated a gating network and a sparse attention mechanism to customize differentiated semantic compression ratios for individual terminals. A joint computing and bandwidth allocation model was formulated to minimize total system energy consumption. The proximal policy optimization (PPO) algorithm was employed to solve the optimization problem. Simulation results demonstrate that

compared with fixed compression ratio schemes

the proposed strategy reduces total system energy consumption by 39%.

关键词

Keywords

references

GÜNDÜZ D , QIN Z , AGUERRI I , et al . Beyond transmitting bits: Context, semantics, and task-oriented communications [J ] . IEEE Journal on Selected Areas in Communications , 2022 , 41 ( 1 ): 5 - 41 .

BOURTSOULATZE E , BURTH KURKA D , GÜNDÜZ D . Deep Joint Source-Channel Coding for Wireless Image Transmission [J ] . IEEE Transactions on Cognitive Communications and Networking , 2019 , 5 ( 3 ): 567 - 579 .

YANG K , WANG S , DAI J , et al . SwinJSCC: Taming Swin Transformer for Deep Joint Source-Channel Coding [J ] . IEEE Transactions on Cognitive Communications and Networking , 2025 , 11 ( 1 ): 90 - 104 .

香晏 , 赵响 , 黄军韬 . 一种基于残差连接的Swin Transformer增强型联合编码架构设计 [J ] . 无线电工程 , 2025 , 55 ( 05 ): 905 - 912 .

XIANG Y , ZHAO X , HUANG J . A Residual Connection-based Swin Transformer Enhanced Joint Encoding Architecture Design [J ] . 无线电工程 , 2025 , 55 ( 05 ): 905 - 912 .

张平 , 牛凯 , 姚圣时 , 等 . 面向未来的语义通信：基本原理与实现方法 [J ] . 通信学报 , 2023 , 44 ( 5 ): 1 - 14 .

ZHANG P , NIU K , YAO S , et al . Semantic communications for future: basic principle and implementation methodology [J ] . Journal on Communications , 2023 , 44 ( 5 ): 1 - 14 .

TANG Q , ZHANG B , LIU J , et al . Dynamic Token Pruning in Plain Vision Transformers for Semantic Segmentation [C ] // 2023 IEEE/CVF International Conference on Computer Vision (ICCV) . Piscataway : IEEE , 2023 : 777 - 786 .

PAN J , BULAT A , TAN F , et al . EdgeViTs: Competing Light-weight CNNs on Mobile Devices with Vision Transformers [C ] // Computer Vision – ECCV 2022 . Cham : Springer , 2022 : 294 - 311 .

GAO Z , TONG Z , WANG L , et al . SparseFormer: Sparse Visual Recognition via Limited Latent Tokens [C ] // International Conference on Representation Learning , 2024 : 55639 - 55657 .

YUAN J , GAO H , DAI D , et al . Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention [C ] // Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , Vienna, Austria, Association for Computational Linguistics , 2025 : 23078 - 23097 .

ZHAO Z , YANG Z , GAN X , et al . A joint communication and computation design for semantic wireless communication with probability graph [J ] . Journal of the Franklin Institute , 2024 , 361 ( 13 ): 107055 .

YANG S , SHEN B , HUANG X . Optimizing Semantic Spectral Efficiency in Wireless Image Transmission: A PPO-Driven Resource Allocation Scheme [J ] . IEEE Communications Letters , 2025 , 29 ( 6 ): 1466 - 1470 .

CANG Y , CHEN M , YANG Z , et al . Online Resource Allocation for Semantic-Aware Edge Computing Systems [J ] . IEEE Internet of Things Journal , 2024 , 11 ( 17 ): 28094 - 28110 .

ZHAO S , LIU T , JIN H , et al . SPViT: Accelerate Vision Transformer Inference on Mobile Devices via Adaptive Splitting and Offloading [J ] . IEEE Transactions on Mobile Computing , 2025 , 24 ( 10 ): 9303 - 9318 .

JI Z , QIN Z , TAO X , et al . Resource Optimization for Semantic-Aware Networks With Task Offloading [J ] . IEEE Transactions on Wireless Communications , 2024 , 23 ( 9 ): 12284 - 12296 .

顾健华 , 冯建华 , 许辉阳 , 等 . 基于有向图与卷积网络强化学习的端侧协同算力资源分配方法 [J ] . 电子学报 , 2025 , 53 ( 06 ): 1771 - 1783 .

GU J , FENG J , XU H , et al . Directed Graph and Convolutional Network Reinforcement Learning for Terminal-Side Collaborative Computing Resource Allocation Scheme [J ] . Acta Electronica Sinica , 2025 , 53 ( 06 ): 1771 - 1783 .

NGUYEN L X , TUN Y L , TUN Y K , et al . Swin Transformer-Based Dynamic Semantic Communication for Multi-User With Different Computing Capacity [J ] . IEEE Transactions on Vehicular Technology , 2024 , 73 ( 6 ): 8957 - 8972 .

ZHANG J , ZHANG Y , JI B , et al . Feature-Driven Semantic Communication for Efficient Image Transmission [J ] . Entropy , 2025 , 27 ( 4 ): 369 .

ALBASEER A , ABDALLAH M , Tailoring Semantic Communication at Network Edge: A Novel Approach Using Dynamic Knowledge Distillation [C ] // ICC 2024 - IEEE International Conference on Communications . Piscataway : IEEE , 2024 : 1455 - 1460 .

ZHENG Y , ZHANG T , MU X , et al . Joint Semantic Transmission and Resource Allocation for Intelligent Computation Task Offloading in MEC Systems [J ] . IEEE Transactions on Wireless Communications , 2025 , 24 ( 10 ): 8756 - 8770 .

LIU C , GUO C , YANG Y , et al . Adaptable Semantic Compression and Resource Allocation for Task-Oriented Communications [J ] . IEEE Transactions on Cognitive Communications and Networking , 2024 , 10 ( 3 ): 769 - 782 .

ZHENG G , WEN M , NING Z , et al . Computation-Aware Offloading for DNN Inference Tasks in Semantic Communication Assisted MEC Systems [J ] . IEEE Transactions on Wireless Communications , 2025 , 24 ( 4 ): 2693 - 2706 .

ZHENG Y , ZHANG T , HUANG R , et al . Joint Computing Offloading and Resource Allocation for Classification Intelligence Tasks in MEC Systems [J ] . IEEE Transactions on Network and Service Management , 2026 , 23 : 1086 - 1099 .

CHEN B , HUANG Y , QIU H , et al . Image Compression for Resource-Constrained AIoT System With Compressed Sensing [J ] . IEEE Transactions on Systems, Man, and Cybernetics: Systems , 2025 , 55 : 5464 - 5476 .

HU H , SONG K , FAN R , et al . Energy-Efficient Image Semantic Communication: Architecture Design and Optimal Joint Allocation of Communication and Computation Resources [J ] . IEEE Transactions on Circuits and Systems for Video Technology , 2025 , 35 ( 7 ): 6466 - 6480 .

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于区域感知的通感算协同网络语义通信可靠性联合优化

可扩展分层次编译码的语义通信系统

高效融合全局和局部上下文特征的语义通信系统

知识增强的语义通信接收端设计

多模态语义通信研究综述