Attack detection in Water Supply Systems using Kalman filter estimator

In this paper we present a framework for the attack and fault detection in a Water Supply System consisting of wireless sensors and drifters. The framework is derived using the Saint-Venant equations for shallow water system. These equations are used to obtain the state space model for Kalman Filter which works as a central estimator. The estimator estimates the sensor readings for the next time cycle based on the readings of the past and neighbor sensors. The estimates from the Kalman filter and the actual observations are then fed to a χ² detector. The detector computes the difference between the two readings and compares it with a given threshold to detect if the system has been compromised. Our discussion shows that the proposed framework can detect various attacks and faults in the system.