Recognizing human activity using deep learning with WiFi CSI and filtering

Abstract

We are living in the era of the Internet of Things, where it is easy to find network access points (APs). APs could be useful for more than just connecting to the Internet. The presence of a human between two APs, as well as human behavior, causes a change in the waveform of a WiFi signal. In a previous research, we have explained how changes in waveforms affect the channel state information of the signal and how machine learning can utilize that information to recognize and predict human behavior. In this paper, we explain the limitation of the last paper and provide a solution for improving the limited performance, which is preprocessing. Kalman filtering improved the training accuracy by 2%. In conclusion, the overall Kalman filter is good for suppressing sudden signal errors such as those from hardware malfunctioning.