Abstract
Federated Learning (FL) algorithms commonly sample a random subset of clients to address the straggler issue and improve communication efficiency. While recent works have proposed various client sampling methods, they have limitations in joint system and data heterogeneity design, which may not align with practical heterogeneous wireless networks. In this work, we advocate a new independent client sampling strategy to minimize the wall-clock training time of FL, while considering data heterogeneity and system heterogeneity in both communication and computation. We first derive a new convergence bound for non-convex loss functions with independent client sampling and then propose an adaptive bandwidth allocation scheme. Furthermore, we propose an efficient independent client sampling algorithm based on the upper bounds on the convergence rounds and the expected per-round training time, to minimize the wall-clock time of FL, while considering both the data and system heterogeneity. Experimental results under practical wireless network settings with real-world prototype demonstrate that the proposed independent sampling scheme substantially outperforms the current best sampling schemes under various training models and datasets.
Original language | English |
---|---|
Title of host publication | ICC 2024 - IEEE International Conference on Communications |
Editors | Matthew Valenti, David Reed, Melissa Torres |
Publisher | IEEE |
Pages | 944-949 |
Number of pages | 6 |
ISBN (Electronic) | 9781728190549 |
ISBN (Print) | 9781728190556 |
DOIs | |
Publication status | Published - 10 Jun 2024 |
Event | 2024 59th IEEE International Conference on Communications, ICC 2024: Scaling the Peaks of Global Communications - Sheraton Denver Downtown Hotel, Denver, United States Duration: 9 Jun 2024 → 13 Jun 2024 https://icc2024.ieee-icc.org/ (Link to conference website) https://icc2024.ieee-icc.org/program (Link to conference programme) |
Publication series
Name | IEEE International Conference on Communications |
---|---|
ISSN (Print) | 1550-3607 |
ISSN (Electronic) | 1938-1883 |
Conference
Conference | 2024 59th IEEE International Conference on Communications, ICC 2024 |
---|---|
Country/Territory | United States |
City | Denver |
Period | 9/06/24 → 13/06/24 |
Internet address |
|
Scopus Subject Areas
- Computer Networks and Communications
- Electrical and Electronic Engineering
User-Defined Keywords
- client sampling
- convergence analysis
- Federated learning
- optimization algorithm
- wireless network