Client Selection in Hierarchical Federated Learning

作     者：Trindade, Silvana da Fonseca, Nelson L. S. 

作者机构：Univ Estadual Campinas Inst Comp BR-13083970 Campinas Brazil 

出 版 物：《IEEE INTERNET OF THINGS JOURNAL》 (IEEE Internet Things J.)

年 卷 期：2024年第11卷第17期

页      面：28480-28495页

核心收录：

学科分类：0810[工学-信息与通信工程] 0808[工学-电气工程] 08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：IBM Fellowship Program, USA CAPES, Brazil CNPq, Brazil [405940/2022-0] CNPq/MCTI, Brazil Sao Paulo Research Foundation (FAPESP), Brazil [2023/00673-7] 

主　　题：Servers Convergence Performance evaluation Training Computational modeling Clustering algorithms Federated learning Client selection edge computing federated learning (FL) hierarchical architecture 

摘      要：Federated learning (FL) is a promising technique for providing distributed learning without clients disclosing their private data. In hierarchical FL (HFL), edge servers partially aggregate the parameters of their connected clients models, improving scalability and reducing computational overhead on the central server. To speed up the convergence of the global model, only those clients with potential contributions to the model performance will participate in the model training. This article introduces a two-step client selection approach for the HFL and three novel algorithms, which consider a large set of features in this selection and the client s contributions to the model performance. Compared to the selected baseline algorithms, the proposed client selection algorithms reduce CPU utilization by more than 50%, memory usage by 80%, and energy consumption by 50%.