ARGUS: Wi-Fi AI Identifies People Without Devices.

Nayan Sanjay Bhatia, Pranay Kocheta, Yuhan Li, Katia Obraczka· August 18, 2026 View original

Key takeaways

  • ARGUS is a passive Wi-Fi system for device-free person identification.
  • It uses Channel State Information (CSI) and attention-guided Transformers.
  • Achieves high accuracy on a large subject dataset with fewer FLOPs.
  • Offers a privacy-preserving alternative to camera/wearable biometrics.

Who benefits

SecuritySmart HomeRetailHealthcare

Summary

Researchers developed ARGUS, a passive Wi-Fi sensing system that identifies individuals using commodity Channel State Information (CSI) without requiring wearables or specific movements. The system employs attention-guided Transformers and achieves high accuracy on a 154-subject dataset, outperforming baselines with fewer computational resources.

A new passive Wi-Fi sensing system, named ARGUS, has been developed for device-free person identification. Unlike traditional biometric methods that rely on cameras or wearables, ARGUS uses commodity Channel State Information (CSI) from standard Wi-Fi networks to identify individuals without requiring them to carry a device or perform specific movements. This system converts short CSI spans into compact "statgrams," which are statistical maps derived from channel views. ARGUS then utilizes a lightweight decoder-only Transformer that processes these statgram patches as tokens, aggregating evidence over time for identification. Evaluated on a dataset of 154 subjects, ARGUS achieved a Top-1 accuracy of 78.88% over 6-second windows, rising to 84.85% when aggregating over 60-second segments. It also reached impressive Top-3 and Top-5 accuracies of 98.61% and 99.26% respectively. Notably, ARGUS improved upon raw-CSI Transformer baselines by 7.75 points while using 4.4 times fewer FLOPs per window, demonstrating its efficiency and scalability.

Why it matters

This technology offers a privacy-preserving and scalable alternative for person identification in various settings, from smart homes to security systems, without the need for intrusive cameras or user compliance with wearables.

How to implement this in your domain

  1. 1Explore ARGUS for passive security monitoring in sensitive areas where cameras are undesirable.
  2. 2Integrate Wi-Fi CSI-based identification into smart home systems for personalized automation.
  3. 3Pilot ARGUS in commercial spaces for anonymous foot traffic analysis and occupancy monitoring.
  4. 4Research ethical implications and privacy safeguards before deploying such identification systems.

Original post by Nayan Sanjay Bhatia, Pranay Kocheta, Yuhan Li, Katia Obraczka

"arXiv:2608.14670v1 Announce Type: new Abstract: Passive, device-free person identification offers an alternative to camera- and wearable-based biometrics, yet existing wireless approaches rely largely on gait or activity cues and are rarely evaluated at scale. In this paper, we p…"

View on X

Originally posted by Nayan Sanjay Bhatia, Pranay Kocheta, Yuhan Li, Katia Obraczka on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses