accuracy

Using WiFi networks as a technology to track people is a twist in the script that not all of us saw coming. He Karlsruher Institute for Technologyone of the strongest research institutions in Germany, assures close to 100% accuracy when recognizing people without any type of camera and using it. What exactly happened. The KIT (Karlsruhe Institute of Technology) team published a paper with a promising headline: “Ordinary WiFi can identify people with almost perfect accuracy”. And this is achieved thanks to something that routers have been doing for recent years: beamforming feedback information. How the hell does this work?. To understand what it is about beamforming You must first understand how the devices emit signals. routers. In their first generations, routers emitted in all directions, just like a light bulb emits light in that way. With the most modern versions of WiFi, the way the signal is transmitted has improved. Routers began to concentrate the signal towards where the receiving device is, like a flashlight instead of a light bulb. Beanformig. That is called beamformingto form a concentrated beam and received by another device. But to aim well, the router needs to know where to point, and it is the connected devices themselves—your cell phone, your laptop—that send that information to the router continuously. Basically, they are constantly telling the router “hey, I’m here.” That message is the BFI, beamforming feedback information. And what is this for?. Now you know that your router sends information to your gadgets and that your gadgets send information to the router. When the devices send information to the router, they describe how the signal arrives, and interference along the way is recorded. Among them, human beings. Our body partially absorbs WiFi waves, reflects them, deflects them and alters how they reach the mobile phone or router. The researchers used that signal data to train models of artificial intelligencein order to detect patterns that would allow humans to be detected. They fed the system with thousands of examples associated with different people until the model learned to detect those wave changes associated with human presence. The system is not capable of visually recognizing anything in the environment, but it manages to have information about when a human is present in the environment. The caution. According to the researchers, “this technology turns each router into a potential means of surveillance.” “If you regularly pass by a café that operates a WiFi network, you could be identified there without realizing it and be recognized later, for example, by public authorities or companies.” The reality? It would be necessary for cybercriminals to develop a system identical or similar to that of the KIT to achieve a human video surveillance system through WiFi signals. The nuance. Under laboratory conditions, with 197 participants and in controlled environments, the system was close to 100% accuracy. But in the real world, it would be necessary to train a new model with data from hundreds of people in different spaces. The model is not a ready-to-deploy technology or a real threat – nor is it intended to be applied – but the research reveals how simple a priori data sets can be trained as a surveillance tool. In Xataka | There is a booming job in the era of artificial intelligence: cybersecurity expert