Wi-Fi signals manipulated to block human activity tracking remotely
Zero-Knowledge Remote Adversarial Attack against Wi-Fi-based Human Activity Recognition for Privacy Protection
Networking and Internet Architecture
Summary
Wi-Fi devices can guess what people are doing just by analyzing signals, which raises privacy worries. The authors introduce a system called GRAW that tricks these devices by changing the Wi-Fi signals coming from the router, making the activity guessing wrong. GRAW works without knowing how the guessing system works and was tested on many types and environments, always making the guesses as bad as random chance. This method also keeps regular Wi-Fi working almost normally and was shown to work live over real airwaves.
What this means in practice
- •For network security teams: Implement protective measures that spoil Wi-Fi based human activity monitoring to safeguard users' privacy.
- •For wireless device manufacturers: Develop routers incorporating signal manipulation techniques to prevent unauthorized activity recognition without disrupting normal communication.$Commercial implications: Allows sale of privacy-enhanced routers that block activity tracking while maintaining Wi-Fi performance.
Authors
Byungjun Kim, Amogh Panchagatti, Peter Gerstoft, Xinyu Zhang, Minsung Kim
Abstract
The growing capability of Wi-Fi devices to identify human activities using channel state information (CSI) raises privacy concerns. To counter this threat, we propose GRAW, an adversary system, acting as a privacy defender, that degrades the human activity recognition (HAR) system at the user device by perturbing the router's signals that the device uses to estimate CSI. GRAW employs generative adversarial imitation learning (GAIL) to construct perturbation signals, and thereby eliminates the need for any information on the target HAR systems and their inputs (i.e., zero-knowledge operation). We evaluate GRAW against seven representative HAR models, using datasets collected in five environments, including our own dataset. We observe that GRAW is the only remote attack scheme that degrades every tested HAR model to a random-selection level. At the same perturbation level, GRAW achieves an attack success ratio up to 76.7% higher than comparison methods, while maintaining over 99% packet success rate on regular Wi-Fi communication. We demonstrate the feasibility of GRAW through real-time, over-the-air experiments with software-defined radios.