1. When Wi-Fi starts to “see” you
Imagine walking through a café without your phone or Wi-Fi, but the wireless signals around you still recognize who you are.
Recent research from the Karlsruhe Institute of Technology (KIT) in Germany shows that Wi-Fi signal feedback alone can identify and track individuals without your device. This technology, known as BFId (Identity Inference Attack), uses Beamforming Feedback Information (BFI) from Wi-Fi 5 and above to generate radio “images.”
Researchers warn that this capability could make every Wi-Fi router a “silent observer.”
2. BFId: 197 people experiment, close to 100% recognition rate
- Research paper: BFId: Identity Inference Attacks Utilizing Beamforming Feedback Information
📄 KIT Official Papers PDF - The research team collected Wi-Fi signal feedback data from 197 participants and trained them at different angles, walking patterns, and movements.
- Using machine learning models to analyze small changes in signals, nearly 100% recognition accuracy is achieved .
- The peculiarity: the target does not need to carry any equipment. As long as the Wi-Fi signal is covered and there is communication activity, you can “see” you.
The researchers noted that current Wi-Fi feedback data is often unencrypted in standard protocols, meaning that any listener could capture this information.
“Every router can become an invisible monitor.”
– Karlsruhe Institute of Technology press release
📎 The Spy Who Came in from the Wi-Fi (KIT Press)
3. WhoFi: “Wireless re-identification” system by the Italian research team
Another study from La Sapienza University in Rome, Italy, proposes a system called WhoFi .
It uses deep learning to identify people through channel state information (CSI) in Wi-Fi signals.
- Papers: WhoFi: Deep Person Re-Identification via Wi-Fi Channel Signal Encoding
- Experimental dataset: NTU-Fi
- Identification accuracy: Rank-1 has an accuracy rate of 95.5%
- It does not rely on cameras, light, or equipment, and can even be recognized through walls.
“Everyone’s way of blocking Wi-Fi is unique.”
— PCPer Technology Media
📎 Report link
4. Privacy crisis: “non-inductive surveillance” from camera to signal
Two studies reveal the same trend:
Future surveillance may not require cameras.
| Technology | Use signals | Identification accuracy | Whether you need to bring equipment | Potential risks |
|---|---|---|---|---|
| BFId (KIT) | Beamforming Feedback BFI | ≈100% | Nope | No device recognition / spatial tracking |
| WhoFi (Sapienza) | Channel Status Information CSI | ≈95% | Nope | Wall-through recognition / no visual monitoring |
This means that as long as the Wi-Fi signal exists, the “wireless fingerprint” of human activity can be recorded, analyzed, and re-identified.
The IEEE is currently advancing the 802.11bf Wi-Fi perception standard, formalizing such perception capabilities into the Wi-Fi technology framework.
👉 Interesting Engineering reports
5. What can we do?
- Encrypt or obfuscate BFI/CSI data in Wi-Fi protocols.
- Introducing privacy protection requirements into standard development.
- Ordinary users should understand that turning off Bluetooth and Wi-Fi is not foolproof, and true security depends on the privacy protection of the device layer and the protocol layer.
“The future of Wi-Fi Sensing is full of potential, but without regulation, it could also become the most discreet way to monitor.”
References
- J. Todt et al., BFId: Identity Inference Attacks Utilizing Beamforming Feedback Information, Karlsruhe Institute of Technology, 2025.
- D. Avola et al., WhoFi: Deep Person Re-Identification via Wi-Fi Channel Signal Encoding, arXiv preprint, 2025.
- The Spy Who Came in from the Wi-Fi, KIT Press Release, 2025-10-08.
- TechRadar, PCPer, TechXplore Coverage Collection, July – October 2025.
Author’s Note: This article is based on public papers and institutional reports, and aims to promote public understanding of wireless perception and privacy issues.
Tubing: