Occupancy Estimation Using Wifi Motion Detection via Supervised Machine Learning Algorithms

WiFi signals have tendency of getting disturbed by the motion of occupants and other movements in a zone. If we measure the level of this variation, it can represent the human activity in that zone. In this paper, we have proposed estimation of the occupancy by classifying the activity level obtained by disturbance in WiFi signal using several supervised machine learning approaches. We have prepared class labels using the schedule of people in the zone and verified it by counting the number of persons each hour. The proposed framework is tested and validated by collecting the data from an office space in a building and different performance measures are computed to see the effectiveness of this framework in occupancy estimation. In this task, Decision Tree and Random Forest are most stable with the highest accuracy of 95%.