Adversary-Aware Private Inference over Wireless Channels

TL;DR

This paper introduces a new framework for privacy-preserving AI inference over wireless channels by transforming features before transmission, addressing individual feature privacy in edge sensing scenarios.

Contribution

It proposes a novel feature transformation framework to protect individual data privacy during wireless feature transmission for AI inference.

Findings

Effective feature transformation reduces privacy risks.

Framework maintains inference accuracy while enhancing privacy.

Addresses a gap in privacy protection for individual features.

Abstract

AI-based sensing at wireless edge devices has the potential to significantly enhance Artificial Intelligence (AI) applications, particularly for vision and perception tasks such as in autonomous driving and environmental monitoring. AI systems rely both on efficient model learning and inference. In the inference phase, features extracted from sensing data are utilized for prediction tasks (e.g., classification or regression). In edge networks, sensors and model servers are often not co-located, which requires communication of features. As sensitive personal data can be reconstructed by an adversary, transformation of the features are required to reduce the risk of privacy violations. While differential privacy mechanisms provide a means of protecting finite datasets, protection of individual features has not been addressed. In this paper, we propose a novel framework for privacy-preserving AI-based sensing, where devices apply transformations of extracted features before transmission to a model server.