科研速览 · Science Skim继续刷下去 · Keep skimming →
◆ Proceedings of the ACM on Interactive Mobile Wearable and Ubiquitous Technologies2026-03-16· Obfuscation

MetAegis: Defend Against Physical-layer Wireless Sensing Leakage via Metasurface Obfuscation

Jinjiang Yang, Chenghan Jiang, Jie Zhang, Xinyi Li, Boxuan Yang, Shibo Zhang, Xianjia Meng, Zhuzhu Wang, Dingyi Fang, Zhanyong Tang

原始摘要(英文原文)· Original abstract
Wi-Fi-based non-intrusive sensing has emerged as a promising technology for monitoring human activities without the need for wearable devices, attracting significant interest from both industry and academia. However, the inherent broadcast nature of wireless signals renders user activity features vulnerable to eavesdropping by malicious entities, posing serious privacy risks. Existing privacy protection strategies predominantly rely on source-side solutions, such as hardware modifications to encrypt wireless channels. While partially effective, these approaches often necessitate costly and impractical hardware upgrades and, in some cases, fail to accommodate the needs of authorized users, thereby compromising their sensing performance in practical scenarios. To overcome these limitations, we propose MetAegis, an innovative channel-side obfuscation system that leverages a programmable metasurface for flexible and cost-effective deployment. MetAegis dynamically alters the wireless channel to achieve selective feature obfuscation across multiple directions, effectively shielding against malicious users while ensuring that authorized users experience no degradation in performance. This is accomplished through a feature obfuscation technique driven by switching coding configurations on the metasurface, coupled with a feature recovery framework that enables authorized users to accurately reconstruct activity-related features from obfuscated signals. We implemented MetAegis using off-the-shelf Wi-Fi devices and conducted extensive evaluations across three real-world scenarios. Experimental results demonstrate that MetAegis reduces eavesdropper recognition accuracy to below 24% on average, while authorized users maintain average accuracy exceeding 88% through the recovery mechanism under the regular deployment. Remarkably, unlike traditional source-side solutions that require close integration with transmitters, MetAegis sustains its effectiveness even under challenging conditions, such as when the metasurface is positioned over 5 meters from the transmitter. These findings underscore the practicality and robustness of MetAegis in safeguarding privacy without sacrificing usability.
读原文 · Read the paper ↗

AI 追问PRO

登录后使用 AI 追问

讨论区

登录后参与讨论

相关论文 · Related

MetAegis: Defend Against Physical-layer Wireless Sensing Leakage via Metasurface Obfuscation — 科研速览 Science Skim