Countering radiometric signature exploitation using adversarial machine learning based protocol switching

Abstract

A Radiometric signature refers to transceiver specific features that are caused by variations in the manufacturing process even for the same circuit design. While such a radiometric signature constitutes a fingerprint that can be exploited for device authentication, it is a threat to privacy. Particularly, in the realm of wireless networks, an adversary may exploit radio frequency (RF) fingerprinting to identify devices and conduct traffic analysis in order to uncover the topology and categorize the role of various nodes. In this paper, we show how an adversary could employ RF fingerprinting to distinguish among nodes and bypass the provisioned anonymity protection in the network. We analyze the accuracy of RF fingerprinting and highlight how the accuracy affects the success of adversary attacks. To counter such a threat, we propose a novel methodology that requires no hardware changes to the radio transceiver and the associated host device. Our methodology is based on coordinated switching among preset link-layer and physical-layer communication protocols. For the latter, we particularly exploit distributed beamforming. We employ adversarial machine learning to select the protocol configuration for each transmission so that the accuracy of the RF fingerprinting diminishes. We demonstrate the effectiveness of our scheme through simulation and prototype experiments.