anomaly detection in non-stationary water distribution grids using fog computing architecture

efficient monitoring and quick feedback control are the main requirements of smart cities to guarantee the stability and safety of urban infrastructures. real-time monitoring in order to detect anomalies leads to the intensive data processing and hence requires a new computing scheme to offer large-scale and low latency services. fog architecture by extending computing to the edge of the network, provides the ability to accurate and fast detection of abnormal patterns. a hierarchical fog computing architecture and an efficient hyperellipsoidal clustering algorithm presented in previous studies have been applied to identify anomalous behaviors in water distribution grids. we created an urban water distribution grid dataset using epanet2w simulator software by measuring grid features: pressure and head for several scenarios. we created 12 distinct events (unexpected behavior) with different scales during the simulation time. to evaluate the effectiveness of the hierarchical anomaly detection model in water distribution grids, the data and computing nodes at different layers were executed as docker containers. the evaluation results proved the efficiency of the proposed hierarchical anomaly detection model with a significant reduction in latency compared to the centralized scheme, while reaching a significant detection accuracy compared to the centralized one.