DoCoFL: Downlink Compression for Cross-Device Federated Learning

TL;DR

This paper introduces DoCoFL, a novel downlink compression framework for cross-device federated learning, effectively reducing bandwidth usage while maintaining model accuracy, and compatible with uplink compression schemes.

Contribution

DoCoFL is the first framework specifically designed for downlink compression in cross-device federated learning, enabling bi-directional bandwidth reduction.

Findings

Significant bi-directional bandwidth savings achieved.

Maintains competitive model accuracy with no compression baseline.

Compatible with various uplink compression methods.

Abstract

Many compression techniques have been proposed to reduce the communication overhead of Federated Learning training procedures. However, these are typically designed for compressing model updates, which are expected to decay throughout training. As a result, such methods are inapplicable to downlink (i.e., from the parameter server to clients) compression in the cross-device setting, where heterogeneous clients $\textit{may appear only once}$ during training and thus must download the model parameters. Accordingly, we propose $\textsf{DoCoFL}$ -- a new framework for downlink compression in the cross-device setting. Importantly, $\textsf{DoCoFL}$ can be seamlessly combined with many uplink compression schemes, rendering it suitable for bi-directional compression. Through extensive evaluation, we show that $\textsf{DoCoFL}$ offers significant bi-directional bandwidth reduction while achieving competitive accuracy to that of a baseline without any compression.