Physics-Informed Neural Networks for Securing Water Distribution Systems
This work addresses security vulnerabilities in critical infrastructure like water networks, though it appears incremental as it applies an existing PINN method to a new domain.
The paper tackled securing water distribution systems against cyberattacks by applying physics-informed neural networks (PINNs) to mitigate controller attacks, demonstrating a proof-of-concept scenario with potential for enhanced security in cyberphysical systems.
Physics-informed neural networks (PINNs) is an emerging category of neural networks which can be trained to solve supervised learning tasks while taking into consideration given laws of physics described by general nonlinear partial differential equations. PINNs demonstrate promising characteristics such as performance and accuracy using minimal amount of data for training, utilized to accurately represent the physical properties of a system's dynamic environment. In this work, we employ the emerging paradigm of PINNs to demonstrate their potential in enhancing the security of intelligent cyberphysical systems. In particular, we present a proof-of-concept scenario using the use case of water distribution networks, which involves an attack on a controller in charge of regulating a liquid pump through liquid flow sensor measurements. PINNs are used to mitigate the effects of the attack while demonstrating the applicability and challenges of the approach.