Water Quality Evaluation Through Channel State Information in WiFi Networks

Checking fresh water quality cannot be done by using our naked eyes. People always use sensors to evaluate water quality. Here in Iraq, people cannot drink tank water which supplied by local water plant systems since it is not pure. In this paper, we implemented a WiFi based sensor to classify pure versus non-pure water. The sensor was built using the ubiquitous WLAN and Intel 5300 NIC with its modified firmware. The sensor uses the extracted Channel State Information (CSI) of the received WiFi signals, along with Machine Learning (ML) approaches in order to perform the classification process. Promising results were obtained, where the observed accuracies reached up to 98.8% in differentiating pure water from tank water. This paves the road for new type of water quality sensors which characterized by lower cost, less effort, and easier to implement.