Human identification using wifi signal

Abstract

There have been a large number of methods already exists to identify human(e.g.,face recognition, gait recognition, fingerprint identification, etc.). Channel State Information(CSI) obtained from Wifi chipsets already has proven to be a efficient for detecting humans uniquely. We are presenting a system which can identify human uniquely and we are showing that Wifi signal can be used for identifying humans. We are working on the channel properties of a communication link which describes how a signal propagates from the transmitter to receiver and represents the combined effect. Each of the individuals have unique gait and also it is proven. Therefore, for that every human would have distract signal uniquely in the same Wifi spectrum. Our system will analysis the Channel State Information(CSI) to acquire unique features of an individual which will allow us to identify a human precisely. We have used two separate algorithms with an accuracy of 95% to 84% in Decision Tree and 97.5% to 78% in Random Forest between a group of 2 to 5 people. We propose that this technology can be used in office or in smart homes for security reasons as it is allowing us to identify humans.