FreeSense: identifying a person to distort a wireless signal

High technology has long been part of our lives. The concept of a smart home no longer seems something unusual and surprising, such systems work in tens of millions of apartments and houses. If not the whole system, then some part of it. In order to make it all more convenient, the developers are trying to enhance the interaction of technology and man, to make the devices recognize the habits and behaviors of the person and adapt accordingly.

So there are smart lamps and smart refrigerators that know when to turn on or turn off the lights and what kind of food you need to order from the store. In order for such smart systems to immediately identify the owner, developers create on-the-fly identification systems. It uses various kinds of cameras, voice spectrum analyzers, fingerprint sensors. In the near future, another system may appear in homes and offices that can identify a person by the distortion of the wireless network signal. Distortions are caused by the human body, the changes are analyzed and compared with the “control measurement”.

The developers of this system are specialists from the North-West Polytechnic University. According to them , a working prototype has already been created, the accuracy of its work is about 90%. This is not too much, but for the proof of concept the systems are pretty good. The project was named FreeSense. The system of perturbation of the radio signal determines the parameters of the human body, which appeared in the coverage area of ​​the wireless network, as well as features of the body movement that are unique to each of us.
')
The system takes information for analysis from the CSI (channel state information) of the wireless network. “Due to the fact that the characteristics of each person’s body are unique, as well as the movements are unique, any person can be identified by these characteristics. You have just entered the premises, and FreeSense has already identified you, ”says one of the participants in the study. In this case, the person is not required to perform any additional actions. In other cases, you need to enter a code, put your finger on the sensor, or say a couple of words (code or normal). This is not always convenient.

In order to test the operation of FreeSense, the experts used a regular laptop and WiFi router. The equipment was installed in a room of 30 m ² with ordinary furniture. The experiment agreed to take part 9 volunteers, whose personalities and tried to identify using WiFi CSI.

To begin with, the developers made a "cast" for each volunteer. The person entered the room, and the system analyzed the changes in the wireless network signal. Then all the data was recorded and tied to the identity of the volunteer.

After that, all nine volunteers tested the FreeSense features. People came into the room one by one, and walked around the room in different directions. As it turned out, the accuracy of the system in this case is 75%. This indicator increased with a decrease in the number of volunteers. If the database contained data on only 2 people, the accuracy of identifying each of them increased to 95%. The fact is that for some people the parameters of the body and the characteristics of the movement are very similar. And the more "snapshots" in the database, the higher the likelihood of "twins". According to the authors of the project, the ideal option is 2-6 people in the memory of FreeSense, not more. This option is not suitable for large companies, but it is ideal for home systems such as "smart home". In addition, if the access to the premises in the company has 1-2 people, and they need to be identified, FreeSense can also be used here.

Researchers say that the accuracy of the system can be improved, and now the development team is engaged in improving the technology. In the future, it is planned to add the function of determining one person from several who are simultaneously in the room. In addition, it is planned to study how the increase in the distance between the receiver and the wireless signal transmitter affects the accuracy of the entire system.