WiFi: The Invisible Eye—Ordinary Routers Now Capable of Near-Perfect Identification

Date: August 12, 2026
Source: Karlsruher Institut für Technologie (KIT)

Everyday wireless networks may be evolving into potent surveillance instruments. Recent research indicates that standard WiFi signals can be leveraged to track and identify individuals without the need for cameras, specialized sensors, or even a device carried by the subject.

The Mechanism: Radio-Based Imaging

Researchers have demonstrated that the unencrypted signals typically exchanged between routers and devices can be used to construct radio-frequency images of people, allowing for identification within a matter of seconds.

"By observing the propagation of radio waves, we can create an image of the surroundings and of persons who are present," explains Professor Thorsten Strufe from KASTEL (KIT's Institute of Information Security and Dependability).

Essentially, this process mirrors how a traditional camera functions, with one critical distinction: it utilizes radio waves instead of light waves. Because the system analyzes how these waves move through a physical space, the target does not need to possess a smartphone, smartwatch, or any other WiFi-enabled hardware.

You must be connected to the network to be tracked. In reality, it is sufficient that other devices in the vicinity are active.

Signal Propagation Logic

The physics of this identification can be conceptualized by the way waves interact with objects. If we consider the distance $d$ and the velocity of the signal $v$, the time $t$ for a reflection is: $t = \frac{2d}{v}$ By analyzing these reflections from multiple angles, a high-resolution "image" of the person is formed.

Technical Breakdown: Why This is Different

Previous attempts at wireless sensing often required expensive or niche hardware. The new method is revolutionary because it uses existing infrastructure.

The system relies on Beamforming Feedback Information (BFI). These are signals that connected devices routinely send to routers to optimize the connection. Crucially, BFI is transmitted without encryption, making it accessible to anyone within range.

Experimental Results and Accuracy

The research team conducted a rigorous study to validate the system:

• Participants: 197 individuals.
• Accuracy: Nearly 100%.
• Variables: The system remained accurate regardless of the angle of approach or the individual's gait.

Once the machine learning model is trained to recognize a specific person, the identification occurs in just a few seconds.

The Privacy Implications

The omnipresence of WiFi in cafes, offices, and homes creates a massive, invisible surveillance grid. Julian Todt of KASTEL warns that this turns every router into a potential spy tool. For instance, a person passing a cafe could be identified and logged without their knowledge, providing data to corporations or government agencies.

While Felix Morsbach notes that intelligence agencies already have access to CCTV and smart doorbells, the danger here is the invisibility of the medium. Unlike a camera, a router provides no visual cue that it is monitoring you.

Potential Risks and Concerns

• Authoritarian Abuse: Monitoring political protesters without visible security.
• Corporate Tracking: Mapping consumer movement in retail spaces without consent.
• Lack of Consent: Identification occurring without the subject carrying a device.

Conclusion and Call to Action

Professor Strufe emphasizes that while the technology is impressive, it poses a significant threat to fundamental privacy rights. The researchers argue that because wireless networks are already ubiquitous, we cannot wait for the technology to be exploited at scale.

def privacy_recommendation():
    action = "Integrate privacy safeguards into future WiFi standards"
    urgency = "High"
    return f"Recommendation: {action} | Urgency: {urgency}"

print(privacy_recommendation())

The team is calling for the immediate development of protective measures to be baked into the hardware and protocols of future wireless technologies.