Federated Learning over Next-Generation Ethernet Passive Optical Networks

TL;DR

This paper proposes a dynamic wavelength and bandwidth allocation algorithm to enhance QoS for federated learning traffic over next-generation Ethernet Passive Optical Networks, reducing delays and prioritizing FL data.

Contribution

It introduces a novel bandwidth allocation algorithm specifically designed for federated learning traffic in high-speed optical networks.

Findings

The algorithm effectively prioritizes FL traffic over other data.

It reduces latency for FL and delay-critical applications.

The approach improves QoS in next-generation optical networks.

Abstract

Federated Learning (FL) is a distributed machine learning (ML) type of processing that preserves the privacy of user data, sharing only the parameters of ML models with a common server. The processing of FL requires specific latency and bandwidth demands that need to be fulfilled by the operation of the communication network. This paper introduces a Dynamic Wavelength and Bandwidth Allocation algorithm for Quality of Service (QoS) provisioning for FL traffic over 50 Gb/s Ethernet Passive Optical Networks. The proposed algorithm prioritizes FL traffic and reduces the delay of FL and delay-critical applications supported on the same infrastructure.