Hierarchical over-the-air federated edge learning
File(s)
Accepted version
Author(s)
Aygun, Ozan
Kazemi, Mohammad
Gunduz, Deniz
Duman, Tolga M
Type
Conference Paper
Abstract
Federated learning (FL) over wireless communication channels, specifically, over-the-air (OTA) model aggregation framework is considered. In OTA wireless setups, the adverse channel effects can be alleviated by increasing the number of receive antennas at the parameter server (PS), which performs model aggregation. However, the performance of OTA FL is severely limited by the presence of mobile users (MUs) located far away from the PS. In this paper, to mitigate this limitation, we propose hierarchical over-the-air federated learning (HOTAFL), which utilizes intermediary servers (IS) to form clusters near MUs. We provide a convergence analysis for the proposed setup, and demonstrate through experimental results that local aggregation in each cluster before global aggregation leads to a better performance and faster convergence than OTA FL.
Date Issued
2022-08-11
Date Acceptance
2022-05-01
Citation
ICC 2022 - IEEE International Conference on Communications, 2022, pp.3376-3381
ISSN
1550-3607
Publisher
IEEE
Start Page
3376
End Page
3381
Journal / Book Title
ICC 2022 - IEEE International Conference on Communications
Volume
2022-May
Copyright Statement
Copyright © 2022 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.
Source
IEEE International Conference on Communications (ICC)
Subjects
clustering
hierarchical federated learning
machine learning
over-the-air communication
Science & Technology
Technology
Telecommunications
Publication Status
Published
Start Date
2022-05-16
Finish Date
2022-05-20
Coverage Spatial
Seoul, South Korea