面向万物智联的语义通信网络

doi:10.11959/j.issn.2096-3750.2021.00209

物联网学报 ›› 2021, Vol. 5 ›› Issue (2): 26-36.doi: 10.11959/j.issn.2096-3750.2021.00209

• 专题：物联网边缘智能与雾计算技术 • 上一篇下一篇

面向万物智联的语义通信网络

石光明¹^,³, 肖泳²^,³, 李莹玉², 高大化¹^,³, 谢雪梅¹^,³

¹ 西安电子科技大学人工智能学院，陕西西安 710071
² 华中科技大学电子信息与通信学院，湖北武汉 430074
³ 琶洲实验室，广东广州 510330

修回日期:2021-02-04 出版日期:2021-06-30 发布日期:2021-06-01
作者简介:石光明（1965- ），男，博士，西安电子科技大学教授、博士生导师，人工智能学院领军教授，教育部长江学者特聘教授，享受国务院政府特殊津贴专家，IEEE/IET Fellow，中国电子学会会士，人工智能学会高级会员， 2017年以第一完成人荣获国家自然科学奖二等奖1项，主要研究方向为语义通信、类脑感知、压缩感知、计算成像、机器学习、脑电信息处理与脑启发智能技术等
肖泳（1980- ），男，博士，华中科技大学教授，IMT-2030 (6G) 推进组网络智能方向副组长，5G联创行业应用开发实验室副主任，IEEE高级会员，中国通信学会高级会员，IEEE Transactions on Mobile Computing副编辑，主要研究方向为网络人工智能、边缘计算、通信网络博弈理论等
李莹玉（1991- ），女，博士，华中科技大学电子信息与通信学院在站博士后，主要研究方向为物联网、网络大数据分析、分布式优化理论等
高大化（1979- ），男，博士，西安电子科技大学人工智能学院教授、博士生导师，陕西省“青年科技新星”，主持国家重点研发计划、国家自然科学基金等项目10余项，发表智能计算成像相关论文30余篇，授权国际、国内专利10余项，主要研究方向为智能计算成像、智能信息处理等
谢雪梅（1967- ），女，博士，西安电子科技大学人工智能学院教授、博士生导师， IEEE高级会员，入选教育部“新世纪优秀人才支持计划”，获得第十八届中国科协年会全国科技工作者创新创业大赛陕西赛区银奖，在国际期刊和会议上发表了 100多篇学术论文，主要研究方向为人类动作识别、目标检测、场景理解、视频分析、深度学习和特征表示等
基金资助:
国家自然科学基金资助项目(61621005);国家自然科学基金资助项目(61632019);国家自然科学基金资助项目(61836008);国家自然科学基金资助项目(62071193);中国博士后科学基金会面上项目(2020M672357)

Semantic communication networking for the intelligence of everything

Guangming SHI¹^,³, Yong XIAO²^,³, Yingyu LI², Dahua GAO¹^,³, Xuemei XIE¹^,³

¹ School of AI, Xidian University, Xi’an 710071, China
² School of Electronic Information and Communications, Huazhong University of Science and Technology, Wuhan 430074, China
³ Pazhou Lab, Guangzhou 510330, China

Revised:2021-02-04 Online:2021-06-30 Published:2021-06-01
Supported by:
The National Natural Science Foundation of China(61621005);The National Natural Science Foundation of China(61632019);The National Natural Science Foundation of China(61836008);The National Natural Science Foundation of China(62071193);China Postdocotoral Science Foundation(2020M672357)

摘要/Abstract

摘要：

随着无线通信智能化应用需求的快速提升，未来通信网络将从单纯追求高传输速率的传统架构向面向万物智联的全新架构转变。语义通信是一种可将用户的需求和信息含义融入通信过程中的全新架构，该架构有望成为未来万物智联网络的新型基础范式。介绍了语义通信与万物智联网络的关系及语义通信的基本模型和组成。通过分析点对点语义通信的局限性，提出以知识共享和资源融合为基础的语义通信网络将更适合成为大规模智联网络的基础；通过分析语义通信网络的基本组成，介绍了一种基于联邦边缘智能的语义通信网络范例。仿真结果显示，语义通信网络将有望在保证数据安全的同时，大幅度降低通信的资源需求量并提高通信效率。最后，探讨了语义通信网络未来发展的开放性问题。

关键词: 语义通信, 万物智联, 边缘智能, 联邦学习

Abstract:

With the fast growing demand for the wireless network intelligence, the future communication system will transform from the traditional data-oriented solution to a novel intelligence-of-everything (IoE)-based architecture.Semantic communication is a new communication technology which involves the meaning of message into the communication process.It is believed that semantic communication will have the potential to serve as the fundamental paradigm for the future IoE.The relationship between the semantic communication and IoE was discussed and the basic models and fundamental components of semantic communication were introduced.By discussing the limitations of point-to-point semantic communication, it was argued that the knowledge sharing and resource convergence-based semantic communication networking would be ideal for supporting the future massive scales of IoE systems.The basic components of the semantic communication networking system were discussed and a federated edge intelligence-based semantic communication networking architecture as a case study was considered.Simulation results show that semantic communication networking has the potential to further reduce the resource demand and improve the efficiency of semantic communication.Finally, open problems for future research were discussed.

Key words: semantic communication, intelligence of everything, edge intelligence, federated learning

中图分类号:

TN929.5

石光明, 肖泳, 李莹玉, 高大化, 谢雪梅. 面向万物智联的语义通信网络[J]. 物联网学报, 2021, 5(2): 26-36.

Guangming SHI, Yong XIAO, Yingyu LI, Dahua GAO, Xuemei XIE. Semantic communication networking for the intelligence of everything[J]. Chinese Journal on Internet of Things, 2021, 5(2): 26-36.

图/表 5

表1

图1

表2

图2

表3

参考文献 26

[1]	XIAO Y , SHI G M , LI Y Y ,et al. Toward self-learning edge intelligence in 6G[J]. IEEE Communications Magazine, 2020,58(12): 34-40.
[2]	ITU-T. Architectural framework for machine learning in future net-works including IMT-2020:Recommendation ITU-T Y.3172[S]. 2019.
[3]	石光明, 李莹玉, 谢雪梅 . 语义通讯:智能时代的产物[J]. 模式识别与人工智能, 2018,31(1): 91-99.
	SHI G M , LI Y Y , XIE X M . Semantic communications:outcome of the intelligence era[J]. Pattern Recognition and Artificial Intelligence, 2018,31(1): 91-99.
[4]	XIAO Y , HIRZALLAH M , KRUNZ M . Distributed resource allocation for network slicing over licensed and unlicensed bands[J]. IEEE Journal on Selected Areas in Communications, 2018,36(10): 2260-2274.
[5]	XIAO Y , KRUNZ M . Distributed optimization for energy-efficient fog computing in the tactile Internet[J]. IEEE Journal on Selected Areas in Communications, 2018,36(11): 2390-2400.
[6]	HOCKETT C F , SHANNON C L , WEAVER W . The mathematical theory of communication[J]. Language, 1953,29(1): 69.
[7]	COVER T M , THOMAS J A . Elements of information theory[M]. New York,USA: John Wiley ＆ Sons,Inc., 1991.
[8]	SHANNON C E . A mathematical theory of communication[J]. The Bell System Technical Journal, 1948,27(3): 379-423.
[9]	CARNAP R , BAR-HILLEL Y . An outline of a theory of semantic information:RLE (Research Laboratory of Electronics) Technical Reports 247[R]. Massachusetts Institute of Technology,Cambridge MA, 1952.
[10]	FLORIDI L . Outline of a theory of strongly semantic information[J]. Minds and Machines, 2004,14(2): 197-221.
[11]	BAO J , BASU P , DEAN M K ,et al. Towards a theory of semantic communication[C]// Proceedings of 2011 IEEE Network Science Workshop. Piscataway:IEEE Press, 2011: 110-117.
[12]	石光明 . 语义通信变革现代通信模式[J]. 战略前沿技术, 2020.
	SHI G M . Semantic communication transforms modern communication patterns[J]. Strategic Frontier Technology, 2020.
[13]	SHI G M , XIAO Y , LI Y Y ,et al. From semantic communication to semantic-aware networking:model,architecture,and open problems[J]. arXiv:2012.15405, 2020.
[14]	钟义信, 张瑞 . 信息生态学与语义信息论[J]. 图书情报知识, 2017(6): 4-11.
	ZHONG Y X , ZHANG R . Information ecology and semantic information theory[J]. Documentation,Information ＆ Knowledge, 2017(6): 4-11.
[15]	钟义信 . 信息理论的新阶段:语义信息论[Z]. 2016.
	ZHONG Y X . The new stage for information theory:semantic infor-mation theory[Z]. 2016.
[16]	张平 . 移动通信堆叠处理模式难以为继 6G 亟需原创理论创新[Z]. 2020.
	ZHANG P . Challenges in traditional stacking-based processing mode urging original theoretical innovation in 6G[Z]. 2020.
[17]	JUBA B . Universal semantic communication[M]. Berlin: Springer, 2011.
[18]	JUBA B , SUDAN M . Universal semantic communication[C]// Proceedings of the 40th annual ACM symposium on Theory of computing. New York:ACM Press, 2008.
[19]	B JUBA , M SUDAN . Universal semantic communication II:a theory of goal-oriented communication[C]// Electronic Colloquium on Computational Complexity (ECCC) TR08-095.[S.l.:s.n.], 2008.
[20]	GüLER B , YENER A , SWAMI A . The semantic communication game[J]. IEEE Transactions on Cognitive Communications and Networking, 2018,4(4): 787-802.
[21]	SCHWARTZ R , DODGE J , SMITH N A ,et al. Green AI[J]. Communications of the ACM, 2020,63(12): 54-63.
[22]	XIAO Y , LI Y , SHI G ,et al. Optimizing resource-efficiency for federated edge intelligence in IoT networks[C]// Proceedings of International Conference on Wireless Communications and Signal Processing.[S.l.:s.n.], 2020: 86-92.
[23]	ZHOU Z , CHEN X , LI E ,et al. Edge intelligence:paving the last Mile of artificial intelligence with edge computing[J]. Proceedings of the IEEE, 2019,107(8): 1738-1762.
[24]	MCMAHAN B , MOORE E , RAMAGE D ,et al. Communication-efficient learning of deep networks from decentralized data[C]// Artificial Intelligence and Statistics. PMLR, 2017: 1273-1282.
[25]	KIPF T N , WELLING M . Semi-supervised classification with graph convolutional networks[EB]. 2016
[26]	XIAO Y , KRUNZ M . Dynamic network slicing for scalable fog computing systems with energy harvesting[J]. IEEE Journal on Selected Areas in Communications, 2018,36(12): 2640-2654.

数据集名称	网络模型	本地存储/KB	有标签数据量/个	目标精度	模型训练耗时/s
MNIST	2层卷积层	46 023	48 000	98.8%	344
MNIST	1层全连接层	46 023	48 000	98.8%	344
Fashion-MNIST	2层卷积层	46 092	48 000	87.0%	369
Fashion-MNIST	1层全连接层	46 092	48 000	87.0%	369
	2层卷积层
CIFAR-10	2层全连接层	121 475	40 000	62.0%	151
	1层线性变换层

数据集	目标精度	网络模型	协同边缘服务器数/个	本地存储/KB	所需标签数据量/个	模型训练耗时/s
CIFAR-10	62.0%	2层卷积层	1	121 475	40 000	151
		2层全连接层	4	30 551	10 000	62
		1层线性变换层	8	15 397	5 000	44
MNIST	98.8%	2层卷积层	1	46 023	48 000	344
		1层全连接层	4	11 570	12 000	126
			8	5 828	6 000	80
FashionMNIST	87.0%	2层卷积层	1	46 092	48 000	369
		1层全连接层	4	11 688	12 000	133
			8	5 887	6 000	121

数据集	传输标签实体数量
	节点数量	10	100	190	550	1 150	1 750	2 350
Cora	压缩率	0.52%	3.84%	9.38%	20.46%	42.61%	64.77%	86.93%
	错误率	50.22%	23.15%	15.69%	10.93%	7.31%	4.36%	1.29%
	节点数量	10	100	2 000	5 000	7 000	10 000	13 000
Amazon Co-Purchase	压缩率	0.11%	0.77%	14.58%	36.40%	50.94%	72.76%	94.57%
	错误率	62.88%	31.40%	27.22%	20.34%	14.40%	7.54%	1.97%
	节点数量	10	20	40	1 000	10 000	13 000	17 000
Coauthor CS	压缩率	0.065%	0.12%	0.23%	5.46%	54.56%	70.92%	92.74%
	错误率	75.06%	33.35%	19.95%	10.91%	6.97%	4.37%	0.98%

面向万物智联的语义通信网络

Semantic communication networking for the intelligence of everything

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 5

参考文献 26

相关文章 10

Metrics

推荐阅读 0

[1]	耿光磊, 高博, 熊轲, 樊平毅, 陆杨, 王煜炜. 联邦学习赋能6G网络综述[J]. 物联网学报, 2023, 7(2): 50-66.
[2]	江恺, 曹越, 周欢, 任学锋, 朱永东, 林海. 车联网边缘智能：概念、架构、问题、实施和展望[J]. 物联网学报, 2023, 7(1): 37-48.
[3]	郭佳慧, 陈卓越, 高玮, 王玺钧, 孙兴华, 高林. 基于背包模型的联邦学习客户端选择方法[J]. 物联网学报, 2022, 6(4): 158-168.
[4]	郅佳琳, 滕颖蕾, 张新阳, 牛涛, 宋梅. 基于DNN卷积核分割的边缘协作推理性能分析[J]. 物联网学报, 2022, 6(4): 72-81.
[5]	郭英芸, 高博, 张志飞, 张煜, 熊轲. 一种基于带宽分配的联邦学习激励机制[J]. 物联网学报, 2022, 6(4): 82-92.
[6]	陈九九, 郭彩丽, 冯春燕, 刘传宏. 智能网联环境下面向语义通信的资源分配[J]. 物联网学报, 2022, 6(3): 47-57.
[7]	马嘉华, 孙兴华, 夏文超, 王玺钧, 谭洪舟, 朱洪波. 基于标签量信息的联邦学习节点选择算法[J]. 物联网学报, 2021, 5(4): 46-53.
[8]	张建波, 王宇, 聂雪军, 吴国庆, 刘久军, 严俊. 智慧渔业时代的深远海养殖平台控制系统[J]. 物联网学报, 2021, 5(4): 120-136.
[9]	林椿珉, 曾烈康, 陈旭. 边缘智能驱动的高能效无人机自主导航算法研究[J]. 物联网学报, 2021, 5(2): 87-96.
[10]	张彦,张科,曹佳钰. 边缘智能驱动的车联网[J]. 物联网学报, 2018, 2(4): 40-48.