Toward a Smart Real Time Monitoring System for Drinking Water Based on Machine Learning

Drinking-water distribution systems facilitate to carry portable water from water resources such as reservoirs, river, and water tanks to industrial, commercial and residential consumers through complex pipe networks. This system may be affected by acts of pollution that may be intentional or accidental. Hence, it's necessary to prevent any intrusion into water distribution systems and to detect pollution as soon as possible. Therefore, water monitoring is required to maintain a good water quality for human and animal life. In this paper we intend to control the quality of the drinking-water using wireless sensor networks. First, we start with a detailed architecture of our smart system. This architecture uses a new generation of wireless sensors to detect the chemical, physical and microbiological water parameters. After, the water quality limits according to the Tunisian standard will exposed. Then we develop a new detection model of water anomalies. Our model is based on machine learning to detect anomalies and malicious acts in real time. In our solution a data aggregation method is created to minimize the amount of data and reduce the processing time.