Empowering Over-the-Air Personalized Federated Learning via RIS

TL;DR

This paper introduces RIS technology to personalized over-the-air federated learning, effectively addressing data heterogeneity and improving convergence through interference mitigation and tailored aggregation schemes.

Contribution

It presents a novel RIS-enabled personalized AirFL framework with interference elimination and new aggregation schemes for better convergence.

Findings

RIS effectively mitigates interference in personalized AirFL.

Proposed schemes outperform existing baselines in numerical tests.

Enhanced convergence in federated learning with data heterogeneity.

Abstract

Over-the-air computation (AirComp) integrates analog communication with task-oriented computation, serving as a key enabling technique for communication-efficient federated learning (FL) over wireless networks. However, AirComp-enabled FL (AirFL) with a single global consensus model fails to address the data heterogeneity in real-life FL scenarios with non-independent and identically distributed local datasets. In this paper, we introduce reconfigurable intelligent surface (RIS) technology to enable efficient personalized AirFL, mitigating the data heterogeneity issue. First, we achieve statistical interference elimination across different clusters in the personalized AirFL framework via RIS phase shift configuration. Then, we propose two personalized aggregation schemes involving power control and denoising factor design from the perspectives of first- and second-order moments, respectively, to enhance the FL convergence. Numerical results validate the superior performance of our proposed schemes over existing baselines.