Privacy-Preserving Wireless Federated Learning Exploiting Inherent Hardware Impairments

TL;DR

This paper proposes leveraging inherent hardware impairments in wireless federated learning to enhance privacy, using a power allocation strategy that introduces beneficial distortion as a privacy-preserving mechanism within differential privacy guarantees.

Contribution

It introduces a novel approach that exploits hardware-induced distortions as a privacy mechanism in wireless federated learning, with a specific power allocation scheme ensuring differential privacy.

Findings

Hardware impairments can be used to improve privacy in federated learning.

The proposed power allocation scheme effectively balances privacy and model performance.

Numerical results validate the privacy guarantees under various hardware impairment levels.

Abstract

We consider a wireless federated learning system where multiple data holder edge devices collaborate to train a global model via sharing their parameter updates with an honest-but-curious parameter server. We demonstrate that the inherent hardware-induced distortion perturbing the model updates of the edge devices can be exploited as a privacy-preserving mechanism. In particular, we model the distortion as power-dependent additive Gaussian noise and present a power allocation strategy that provides privacy guarantees within the framework of differential privacy. We conduct numerical experiments to evaluate the performance of the proposed power allocation scheme under different levels of hardware impairments.