Contamination detection in drinking water distribution systems using sensor networks

Recent advances in low power networked embedded systems and electrochemical/optical water quality sensors have enabled the generation of low-cost, low-power water quality sensor nodes. These sensor nodes leverage the idea of wireless sensor networks for water quality monitoring in drinking water distribution systems, where large numbers of spatially distributed sensor nodes operate cooperatively to monitor and detect contamination events in a timely manner. However, such approach it is expected to suffer from false alarm errors and missed detection errors due to sensor inaccuracies and modeling uncertainties. This paper aims at developing algorithms for network-wide fusion for processing such inaccurate sensor data in order to minimize detection errors and improve contamination detection probability by correlating imperfect sensor decisions in space and time, taking advantage of the large scale deployment and the drinking water distribution system topology. A simulation example of a simple water distribution system is used to illustrate the proposed methodology and provide intuition through a comparative study.