Scalable Federated Learning over Passive Optical Networks

TL;DR

This paper proposes a scalable federated learning method over passive optical networks that maintains constant bandwidth usage and improves learning accuracy by about 10%.

Contribution

It introduces a two-step aggregation approach enabling scalable federated learning over PONs with constant bandwidth and enhanced accuracy.

Findings

Bandwidth remains constant regardless of client number

Achieves approximately 10% improvement in learning accuracy

Supports scalable federated learning over PONs

Abstract

Two-step aggregation is introduced to facilitate scalable federated learning (SFL) over passive optical networks (PONs). Results reveal that the SFL keeps the required PON upstream bandwidth constant regardless of the number of involved clients, while bringing ~10% learning accuracy improvement.