FlowGuard

FlowGuard is based on the combination of IoT sensing infrastructure and Artificial Intelligence (Graph Neural Networks). Graph Neural Networks (GNNs) and IoT can play a significant role in preventing or mitigating cyber-attacks on public drinking water supply systems. GNNs can analyze large-scale graph data from the water system and identify any anomalous behavior or potential vulnerabilities in the network. GNNs and IoT can help with incident response by quickly identifying the source and extent of the attack, enabling targeted and efficient remediation. By responding quickly and effectively to a cyberattack, panic and potential harm to the general population can be avoided.

The FLOWGUARD project marks a pivotal advancement in water treatment security, reaching its mid–term with remarkable achievements. We’ve successfully gathered and refined extensive historical data on water networks, setting a robust foundation for our cutting–edge analysis. Our team has pioneered the integration of Graph Neural Network (GNN) algorithms to map the intricate network of water distribution, ensuring high fidelity in modeling. Notably, we’ve developed a baseline model employing XGBoost, which has already demonstrated promising potential in detecting anomalies and that will serve as a comparing standpoint for our graph neural networks (GNN) developments. Our current efforts are concentrated on fine–tuning the GNN’s performance, leveraging MLOps strategies to optimize its predictive accuracy. Throughout the process, our milestones have been meticulously documented via the Weights and Biases platform, ensuring transparent progress tracking. Our work not only enhances the security of critical infrastructures like water treament plants but also sets a new standard for utilizing AI in public utility monitoring. Stay tuned for further updates as we continue to push the boundaries of technology for the safety and security of our water systems and the society as a whole.