German researchers used ordinary WiFi signals instead of cameras to identify people; the result turned everyday wireless networks into a powerful new way to recognize individuals with nearly 100% accuracy

A WiFi network usually feels like invisible infrastructure, something that simply moves data from a phone or laptop to the internet. But researchers have shown that the same radio signals already filling homes, offices, cafés and public spaces can reveal far more than whether a device is connected. By studying how WiFi radio waves move through a room, researchers can reconstruct information about people and potentially recognize who is standing there, even when that person carries no connected device. The finding comes from researchers at KASTEL, KIT's Institute of Information Security and Dependability, who demonstrated a system capable of identifying people through ordinary wireless signals. In experiments involving 197 participants, the system reached almost 100 percent identification accuracy. It remained effective when people were viewed from different angles and when they changed the way they walked. The result points to an unusual privacy problem because the technology does not depend on a visible camera or a wearable device.

WiFi Radio Waves can reveal the shape of a person

The basic physics behind the technique is not entirely new. WiFi signals are radio waves, and radio waves do not simply travel from a router to a receiving device along one perfectly straight path. As they move through an environment, they interact with walls, furniture, floors and people. Some of the energy is reflected, scattered or otherwise altered before reaching a receiver.

Those changes contain information about the environment. A person moving through a room becomes part of that radio landscape, changing how the wireless signal propagates. If those changes can be measured precisely enough, a computer can infer where a person is and characteristics of their body or movement. The researchers describe the process as being similar in principle to imaging, except that radio waves replace visible light.


That distinction matters. A conventional camera needs light and a direct optical view of its subject. A radio-based system does not necessarily need either. Radio waves can interact with objects that would obstruct a camera's view, and the resulting measurements can be processed to create a representation of what is happening inside a space.

Professor Thorsten Strufe of KASTEL explained that observing radio-wave propagation makes it possible to create an image of surroundings and the people present. The important point is that the system is not identifying a person because their phone has announced who they are. It is extracting information from the way their presence changes the wireless environment.

You do not need to carry a phone

One of the most unsettling aspects of the research is also one of the easiest to misunderstand. A person does not have to own or carry a WiFi-enabled device for the technique to work.

That is because the system is not necessarily tracking a signal emitted by the individual. Instead, it can analyze signals generated by other devices communicating over the network. If a phone, laptop or another connected device is transmitting nearby, the radio waves it produces can interact with the surrounding environment before being received and analyzed.

In other words, switching off your own smartphone would not automatically make you invisible to this type of system. If other WiFi devices remain active around you, their signals can still interact with your body and the surrounding space. The person's presence becomes part of the physical signal being observed.

This is what separates the approach from many familiar forms of digital tracking. Location systems generally need a device associated with a person. Bluetooth tracking relies on an electronic transmitter. Smartphone-based monitoring requires a phone. A WiFi sensing system can potentially infer the presence and identity of a person from the environment itself.

The Hidden data inside everyday WiFi traffic

The researchers' approach takes advantage of information exchanged during normal wireless communication. Modern WiFi systems use mechanisms that help transmitters direct signals efficiently toward receiving devices. One relevant category is beamforming feedback information, or BFI.

Beamforming allows wireless systems to account for the changing radio environment and improve communication between devices. The feedback contains information about how the wireless signal is behaving as it travels through the surroundings. That information can reveal changes caused by objects and people interacting with the radio field.

The concern raised by the researchers is that such information can be transmitted without encryption in existing wireless systems. If someone within radio range can obtain the relevant information, they may be able to analyze it for purposes beyond ordinary network communication.

This turns an apparently harmless technical exchange into something potentially much more consequential. The information was designed to help wireless devices communicate efficiently, not to identify people. Yet machine-learning systems can learn patterns hidden inside those measurements and associate them with particular individuals.

Machine learning turns radio measurements into identity

The difficult part is not simply detecting that someone is standing in a room. The researchers also wanted to determine who that person was. Machine learning makes that possible by learning recurring patterns in radio measurements associated with different people. Human bodies have different shapes, proportions and movement patterns. Even when two people walk through the same space, they do not necessarily alter the surrounding radio field in exactly the same way.

During training, a recognition model can be exposed to measurements associated with known individuals. Once trained, it can compare new observations with those learned patterns. The result is not a photograph in the conventional sense. It is a radio-derived representation from which the system can extract features useful for recognition.

The researchers tested the approach with 197 participants and reported identification accuracy approaching 100 percent. The system also performed across different perspectives and walking styles, suggesting that the learned signal patterns were not limited to one carefully controlled viewing angle or a single characteristic movement.

That level of accuracy is significant because it moves WiFi sensing beyond simple occupancy detection. Knowing that a person is present in a room is one thing. Determining that the person is a particular individual creates a very different privacy risk.

Why this could be harder to notice than a Camera

Cameras are obvious surveillance devices. They are mounted on walls, doors and ceilings. People can often see them and make a reasonable judgment about where they are being recorded.

Wireless surveillance is different. A router rarely looks like a monitoring device. In most homes and businesses, it is simply part of the infrastructure needed for internet access. Its radio emissions are invisible, and there is no obvious lens pointing toward the people nearby.

Julian Todt of KASTEL warned that this could turn ordinary routers into potential surveillance tools. Someone who regularly walks past a café, for example, might be exposed to the same wireless environment repeatedly without realizing that radio measurements could potentially be used to recognize them.

Felix Morsbach also noted that conventional surveillance methods remain easier to use in many situations. Authorities or criminals may already have access to CCTV cameras or connected video doorbells. The concern is what happens if wireless networks become an additional layer of monitoring that is widespread, difficult to notice and already present in places where people live and work.

A different kind of surveillance infrastructure

The researchers' warning becomes more serious when the technology is considered at scale. WiFi networks are now common in homes, workplaces, shops, hotels, restaurants, transportation facilities and public buildings. Each network creates a radio environment that could potentially be analyzed.

That does not mean every existing router is secretly identifying people. The demonstrated technology depends on specialized analysis, suitable measurements and a trained machine-learning model. There are also practical and technical limitations between a research demonstration and a reliable surveillance system operating across large public spaces.

But the underlying privacy question remains. Wireless networks are already embedded in everyday life, and people generally do not think of them as sensors that can observe physical movement. If their capabilities improve, the distinction between communication infrastructure and sensing infrastructure could become increasingly blurred.

The researchers are especially concerned about authoritarian uses. A system that can identify people without requiring them to carry a device or stand in front of a camera could potentially be used to monitor protesters, visitors or other groups while remaining difficult to detect. The problem is not simply whether the technology works. It is who can deploy it, what information they retain and whether people have any meaningful way to opt out.

Privacy protections may need to be built into WiFi

The researchers argue that these risks should be considered while future wireless standards are being developed. They specifically call for privacy safeguards to be incorporated into the forthcoming IEEE 802.11bf WiFi standard, which is intended to support more advanced WiFi sensing capabilities.

That debate is likely to become more important as wireless networks evolve from systems that primarily transmit information into systems capable of sensing their physical surroundings. The same radio waves that make a connection possible can also carry clues about movement, position and human identity.

The larger lesson is easy to miss. Privacy does not disappear only when someone points a camera at you. It can also erode through ordinary infrastructure operating quietly in the background. WiFi was built to connect devices, but the physics of radio waves gives those networks another capability: they can sense the world around them. The research involving 197 people shows how far that capability can already reach, and it raises a difficult question for the next generation of wireless technology: how much should a network be allowed to know about the people who simply happen to be nearby?