Client Selection for Mobile Edge Computing-Assisted Hierarchical Federated Learning with Non-Independent and Identically Distributed Data
Yi Su, Qingguo Chi, Binbin Shen, Bo Zhang, Jie Zhao, Liang Chen · 2023
This paper focuses on the Mobile Edge Computing-assisted Hierarchical Federated Learning (MEC-HFL) and the Non-Independent and Identically Distributed (Non-IID) data among edges, especially. To improve the learning performance of MEC-HFL systems with fixedly associated edge servers and clients, this paper analyzes the learning process of MEC-HFL and proposes a multi-objective optimization problem which simultaneously minimizes the data distribution distance and maximizes the amount of data through client selection. A Nondominated Sorting Genetic Algorithm-based Client Selection Algorithm (NSGA-CSA) is proposed to solve the above problem. Simulation results show that our proposed NSGA-CSA converges quickly. Under different datasets and models, while improving the learning performance by 2.16%, the NSGA-CSA reduces the learning cost by 21.87%.