Clustering-Based User Selection in Federated Learning: Metadata Exploitation for 3GPP Networks

TL;DR

This paper introduces a metadata-driven federated learning framework that uses a novel data partition model and clustering-based user selection to improve model performance and convergence in realistic, non-IID network scenarios.

Contribution

It proposes a new data partition model based on HPPP and a clustering-based user selection strategy leveraging metadata, addressing data correlation and heterogeneity issues in FL.

Findings

Improves model performance, stability, and convergence in non-IID scenarios.

Maintains comparable performance under IID settings.

Advantages are pronounced with fewer selected users per round.

Abstract

Federated learning (FL) enables collaborative model training without sharing raw user data, but conventional simulations often rely on unrealistic data partitioning and current user selection methods ignore data correlation among users. To address these challenges, this paper proposes a metadatadriven FL framework. We first introduce a novel data partition model based on a homogeneous Poisson point process (HPPP), capturing both heterogeneity in data quantity and natural overlap among user datasets. Building on this model, we develop a clustering-based user selection strategy that leverages metadata, such as user location, to reduce data correlation and enhance label diversity across training rounds. Extensive experiments on FMNIST and CIFAR-10 demonstrate that the proposed framework improves model performance, stability, and convergence in non-IID scenarios, while maintaining comparable performance under IID settings. Furthermore, the method shows pronounced advantages when the number of selected users per round is small. These findings highlight the framework's potential for enhancing FL performance in realistic deployments and guiding future standardization.