Impact Driven Sensor Placement for Leak Detection in Community Water Networks

Community water networks have become increasingly prone to failures due to aging infrastructure, resulting in an increased effort to instrument and monitor networks using IoT (Internet of Things) sensors. However, identifying optimal locations to instrument these sensors to detect and localize failures such as leaks is challenging due to the growing scale and complexity of water networks. Current sensor placement algorithms use heuristics that focus mainly on enabling network coverage. In this paper, we propose a multilevel approach to model and quantify the real-world impact of a failure on a community using various geospatial, infrastructural and societal factors. We present techniques to integrate failure impact, IoT sensing data, and simulation based analytics to drive two novel sensor placement algorithms with the objective of reducing community-scale impact. We evaluate our proposed algorithms on various failure scenarios using multiple real-world water networks at different scales and compare them to existing solutions. The experimental results show that the proposed algorithms result in sensor placements that can achieve an 80% reduction in impact while using a comparable number of sensors for diverse real-world networks.