You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit e254fb9
Browse filesBrowse the repository at this point in the historyBrowse files
author = { Li, Bofan and Ye, Zhankai and Yu, Weikuan and Tang, Yongning and Liu, Xin },
3
+
booktitle = { 2026 IEEE Symposium on Security and Privacy (SP) },
4
+
title = {{ 2FiA: Towards WiFi Sensing-Based Authentication with Unique Biometrics }},
5
+
year = {2026},
6
+
volume = {},
7
+
ISSN = {2375-1207},
8
+
pages = {1618-1634},
9
+
abstract = { The emerging IEEE 802.11bf standard positions WiFi sensing as a key enabler of human biometric sensing for authentication. However, existing WiFi sensing-based authentication systems primarily depend on voluntary or semi-voluntary biometric traits, such as body motions or respiration. While effective, these traits lack strong uniqueness and are susceptible to imitation, making WiFi sensing systems vulnerable to behavioral mimicry attacks. In this paper, we introduce 2FiA the first dual-biometrics WiFi sensing-based authentication system integrating semi-voluntary respiration and involuntary heartbeat. Respiration serves as the primary factor, offering efficient and continuous identity verification, while heartbeat reinforces system robustness through its highly unique and hard-to-imitate nature. For the first time, we propose a novel WiFi sensing pipeline to extract weak heartbeat signals by statistically localizing the thoracic region, effectively suppressing non-thoracic interference, and separating co-located respiration to refine the heartbeat signals. Furthermore, we leverage Respiratory Sinus Arrhythmia (RSA) as a physiological constraint to validate and enhance the extracted heartbeat signals by exploiting the intrinsic coupling between respiration and heartbeat. We then build the authentication model by integrating features from both respiration and heartbeat signals in a unified framework. Through extensive experiments involving dozens of participants across diverse environments, we demonstrate that 2FiA delivers reliable authentication performance, particularly in multi-user and impersonation attack scenarios. },
0 commit comments