A MapReduce based Parallel Niche Genetic Algorithm for contaminant source identification in water distribution network

In recent years, water pollution incidents happen frequently, causing serious disasters and society impact. It is advocated that water quality monitoring sensors shall be deployed in water distribution network to enable real-time pollution detection such that we can effectively detect the water pollution event to reduce the risk. Besides event detection, it is also important to identify the contaminant source for depollution actions. But how to use the information derived from the monitoring sensors to identify the contaminant source is a non-trivial task. Contamination source identification problem is characterized by its extremely high computation complexity, uncertainty and non-uniqueness of the solution in a large-scale water distribution network with dynamic water demands. To tackle this issue, we develop a MapReduce based Parallel Niche Genetic Algorithm (MR-PNGA) that is not only able to achieve high identification accuracy but also to explore the cloud resources for performance improvement. The accuracy and efficiency of MR-PNGA is extensively validated on an 8-server cluster.