Federated Learning With Over-the-Air Aggregation Over Time-Varying Channels

We study federated learning (FL) with over-the-air aggregation over time-varying wireless channels. Independent workers compute local gradients based on their local datasets and send them to a parameter server (PS) through a time-varying multipath fading multiple access channel via orthogonal freque...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on wireless communications 2023-08, Vol.22 (8), p.5671-5684
Main Authors: Tegin, Busra, Duman, Tolga M.
Format: Article
Language:English
Subjects:
Algorithms
Channels
Convergence
doppler spread
Fading
Fading channels
Federated learning
Interference
Machine learning
OFDM
Orthogonal Frequency Division Multiplexing
stochastic gradient descent
Time-varying channels
time-varying multipath fading MAC
Wireless communication
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We study federated learning (FL) with over-the-air aggregation over time-varying wireless channels. Independent workers compute local gradients based on their local datasets and send them to a parameter server (PS) through a time-varying multipath fading multiple access channel via orthogonal frequency-division multiplexing (OFDM). We assume that the workers do not have channel state information, hence the PS employs multiple antennas to alleviate the fading effects. Wireless channel variations result in inter-carrier interference, which has a detrimental effect on the performance of OFDM systems, especially when the channel is rapidly varying. We examine the effects of the channel time variations on the convergence of the FL with over-the-air aggregation, and show that the resulting undesired interference terms have only limited destructive effects, which do not prevent the convergence of the learning algorithm. We also validate our results via extensive simulations, which corroborate the theoretical expectations.
ISSN:1536-1276
1558-2248
DOI:10.1109/TWC.2023.3235894