Experts from the Faculty of Computational Mathematics and Cybernetics of Moscow State University figured out how to most accurately track a person via Wi-Fi. This development can be used in a variety of fields, including transportation, medical, and security systems. All results obtained during the study were published in Proceedings, the scientific publication of RAS’s Institute for Systems Programming.
In general, the need to detect a person and correct his movements can arise in various fields of activity. In total there are two methods associated with this. One is based on the received signal strength indicator, i.e. RSSI, and the other is based on the analysis of more complete information about the status of the Channel State Information. Theoretically, the second method should be more accurate, but it has limitations with the Wi-Fi access points used. The fact is that these access points must be multi-antenna, and the network card of the access point must also have a certain version, and this point itself must have the necessary version of the control software.
According to Andrey Chupakhin, a graduate student and mathematician at Moscow State University, Faculty of Computer Science, Department of Automation of Computing Complex Systems, both of these approaches are based on a similar principle. That is, the RSSI or CSI indicators change as a person switches between Wi-Fi devices. At the same time, using the RSSI-based method, it is possible to get their values in almost all Wi-Fi gadgets.
Experts also suggested two new approaches. In this case, a neural network algorithm (a neural network with repetitive blocks) and a statistical one (Kolmogorov-Wiener filter) are used. The first method turned out to be more accurate.
Source: Ferra
