Humidity Estimation Using WiFi Channel State Information

Abstract

As infrastructure becomes smarter and filled with sensors, the need to combine device functionality is more apparent. Reducing the number of devices by introducing multifunctional sensors will reduce cost and overall complexity. There has been a rapid increase in the development of wireless sensing using WiFi Channel State Information (CSI). WiFi sensing using COTS devices has been a valuable tool in supporting multiple applications such as object movement tracking. One unconventional utility of CSI is to sense meteorological conditions. Temperature, humidity and air pressure are important factors for many applications from air traffic safety to personal health. Given the broad deployment of WiFi, the use of the involved devices in assessing meteorological conditions would enable better coverage and mitigate the cost of employing specialized sensors. This paper empirically demonstrates WiFi sensing, where variations in the CSI are used to estimate humidity using machine learning.