Revealing Hidden IoT Devices through Passive Detection, Fingerprinting, and Localization
Authors: Wei Sun (University of California San Diego), Hadi Givehchian (University of California San Diego), Dinesh Bharadia (University of California San Diego)
Volume: 2025
Issue: 1
Pages: 184–197
DOI: https://doi.org/10.56553/popets-2025-0011
Abstract: Internet-of-things (IoT) devices (e.g., micro camera and microphone) are usually small form factor, low-cost, and low-power, which makes them easy to conceal and deploy in the indoor environment to spy on people for human private information such as location and indoor activities. As a result, these IoT devices introduce a great privacy and ethical threat. Therefore, it is important to reveal these concealed IoT devices in the indoor environment for human privacy protection. This paper presents RFScan, a system that can passively detect, fingerprint, and localize diverse concealed IoT devices in the indoor environment by sensing their unintentional electromagnetic emanations. However, sensing these emanations is challenging due to the weak emanation strength and the interference from the ambient wireless communication signals. To this end, we boost the emanation strength through the non-coherent averaging based on the emanation signal's characteristics and design a novel suppression algorithm to mitigate interference from the wireless communication signals. We further profile emanations across frequency and time that act as the emanation source's unique signature and customize a deep neural network architecture to fingerprint the emanation sources. Furthermore, we can localize the emanation source with an angle-of-arrival (AoA) based triangulation approach. Our experimental results demonstrate the efficiency of the IoT devices' detection, fingerprinting, and localization across different indoor environments.
Keywords: Concealed IoT Devices, Detection, Localization, Fingerprinting, Privacy Enhancement, Wireless Sensing
Copyright in PoPETs articles are held by their authors. This article is published under a Creative Commons Attribution 4.0 license.
