Wi-Fi based supervised machine learning approach to detect objects activities

Abstract

Wireless signal based movement detection technologies are emerging effectively. This type of technology is quite popular compared to other ones because it relieves people from the hassle of wearing sensors and coming in close contact with the detection model. It is possible to detect details of movement pattern of the subject if it comes within the range of wireless signal. In this paper, we have presented a supervised learning approach based on Wi-Fi to detect movements of object. A supervised learning is a machine learning approach where a model is trained with input and desired output beforehand. For our research, we have used a dataset which contains Channel State Information (CSI) data for walk, run, standup, seat etc. separately. In this paper we have worked on three kinds of data such as walk, run and standup. We have trained our model with these CSI data and later applied time series, Augmented Dickey-Fuller Test (ADF) and different machine learning algorithms such as Decision tree, Liner Regression model, Random Forest etc. on those data. Lastly, a rigorous comparison is made between trained and test data in order to validate the accuracy of our result.