PA-iMFL: Communication-Efficient Privacy Amplification Method against Data Reconstruction Attack in Improved Multi-Layer Federated Learning

TL;DR

This paper introduces PA-iMFL, a privacy amplification method for multi-layer federated learning that enhances privacy protection against data reconstruction attacks and reduces communication costs.

Contribution

The paper proposes a novel privacy amplification scheme for improved multi-layer federated learning, combining differential privacy, subsampling, and gradient sign reset to enhance privacy and efficiency.

Findings

PA-iMFL effectively mitigates data reconstruction attacks.

It achieves up to 2.8 times communication efficiency.

Maintains comparable model accuracy to state-of-the-art methods.

Abstract

Recently, big data has seen explosive growth in the Internet of Things (IoT). Multi-layer FL (MFL) based on cloud-edge-end architecture can promote model training efficiency and model accuracy while preserving IoT data privacy. This paper considers an improved MFL, where edge layer devices own private data and can join the training process. iMFL can improve edge resource utilization and also alleviate the strict requirement of end devices, but suffers from the issues of Data Reconstruction Attack (DRA) and unacceptable communication overhead. This paper aims to address these issues with iMFL. We propose a Privacy Amplification scheme on iMFL (PA-iMFL). Differing from standard MFL, we design privacy operations in end and edge devices after local training, including three sequential components, local differential privacy with Laplace mechanism, privacy amplification subsample, and gradient sign reset. Benefitting from privacy operations, PA-iMFL reduces communication overhead and achieves privacy-preserving. Extensive results demonstrate that against State-Of-The-Art (SOTA) DRAs, PA-iMFL can effectively mitigate private data leakage and reach the same level of protection capability as the SOTA defense model. Moreover, due to adopting privacy operations in edge devices, PA-iMFL promotes up to 2.8 times communication efficiency than the SOTA compression method without compromising model accuracy.