面向无线边缘网络的分层Stackelberg博弈群体激励方法  

Hierarchical Stackelberg Game Swarm Learning Incentive Method forWireless Edge Network

作　　者：康海燕[1] 冀珊珊 KANG Hai-yan;JI Shan-shan(School of Information Management,Beijing Information Science and Technology University,Beijing 100192,China)

机构地区：[1]北京信息科技大学信息管理学院,北京100192

出　　处：《电子学报》2024年第7期2382-2392,共11页Acta Electronica Sinica

基　　金：国家社科基金(No.21BTQ079);教育部人文社科基金(No.20YJAZH046);国家自然科学基金(No.61370139)~~。

摘　　要：现有分布式机器学习模型的相关激励机制大多基于单层服务器架构,难以适应当前异构无线计算场景,同时存在计算资源分配不平衡、通信成本高昂等问题.针对上述问题,创新地提出一种面向无线边缘网络的分层Stackelberg博弈群体激励方法(Hierarchical Stackelberg game Swarm Learning Incentive method for wireless edge network,HSISL),创新地将Stackelberg博弈机制引入群体学习模型中,依据各参与方性能差异,云端聚合平台、边缘簇节点、边缘计算节点三方进行动态博弈,通过双定价公平激励过程,共同制定个性化分层资源分配策略,得到模型训练的最优纳什解,有效引导边缘计算模型进行正向加速.通过理论与实验分析,HSISL能够有效提升模型公平性与训练效率,其在MNIST数据集上的准确率可达96.06%.In today’s rapidly evolving landscape of distributed machine learning,conventional data incentive solu⁃tions often fall short due to their reliance on simplistic single-server architectures,in addition,as computing environments become increasingly complex,particularly within the context of heterogeneous wireless networks,these traditional ap⁃proaches struggle to meet the dynamic computational demands such as unbalanced resource allocation and exorbitant com⁃munication costs.In response to the above dilemma,this paper innovatively proposes a hierarchical Stackelberg game swarm learning incentive method for wireless edge network(HSISL).This paper innovatively introduces the Stackelberg game mechanism into the swarm learning.Based on the performance differences of each computing terminal,the cloud ag⁃gregation platform,edge cluster nodes,and edge computing nodes conduct dynamic games and jointly formulate personal⁃ized hierarchical resource allocation strategies through the fair incentive process of dual pricing,which can effectively guide the edge computing model to accelerate forward.Through theoretical and experimental analysis,the HSISL method can ob⁃tain the optimal incentive Nash equilibrium solution for model training.Compared with other incentive methods,the HSISL method can effectively improve the fairness of the model.With training efficiency,its accuracy on the MNIST data set can reach 96.06%.

关 键 词：数据共享 群体学习 STACKELBERG博弈 动态博弈 无线网络通信 激励机制 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术] TN929.52[自动化与计算机技术—计算机科学与技术]