Smart Water Management and Distribution Networks

Abstract

Advances in the Internet of Things and Low-Power Wide-Area Network technologies are driving smart meters to be rolled out within water distribution systems to collect vast quantities of fine-grained data. The challenge is to effectively utilize this large data set to derive actionable insights to limit water loss, and advance efficient and sustainable methods of water distribution. Data availability on the current global scale is limited, but there are widespread acknowledgments of the great potential of utilizing data-centric and machine learning methods, with particular interest and growing emphasis in deep learning for smart water distribution systems specifically, as it captures even more complex structures from large datasets. This work reviews the state of play, notes future challenges, and presents a new taxonomy that classifies water management into three broad categories: infrastructure analysis, demand analysis, and water quality monitoring, while discussing existing ML methods in each category. The work also reviews future research opportunities such as federated learning, incremental learning, probabilistic modelling, and explain-ability as well as some key issues related to data availability and privacy.