Dipartimento di Ingegneria informatica, automatica e gestionale

The optimization and digitalization of Water Distribution Networks (WDNs) are becoming key objectives in our modern society. Indeed, WDNs are typically old, worn and obsolete. These inadequate conditions of the infrastructures lead to significant water loss due to leakages inside pipes, junctions and nodes. It has been measured that in Europe the average value of lost water is about 26 %. Leakage control in current WDNs is typically passive, repairing leaks only when they are visible. Emerging Low Power Wide Area Network (LPWAN) technologies, and especially IoT ones, can help monitor water consumption and automatically detect leakages. In this context, LoRaWAN can be the right way to deploy a smart monitoring system for WDNs. Moreover, most of the current smart WDNs solutions just collect measurements from the smart metres and send the data to the cloud servers, in order to execute the intended analyses, in centralised way. In this paper, we propose new solutions to improve monitoring, leak management and prediction by exploiting edge processing capabilities inside LoRaWAN networks. Our approach is based on an IoT system of water sensors that are placed at junctions of the WDN to have measurements in correspondence to various smart metres in the network and Machine Learning (ML) algorithms to process the data directly at the edge in order to visualise and predict leakages. We present a numerical simulation tool useful to evaluate the suggested monitoring method. Based on our results, we examine whether it is possible to identify network leaks using the edges without having a complete or accurate overview of the collected measurements of the full WDN. System performance is shown separately at gateways network.