A CVAE-Based Anomaly Detection Algorithm for Cyber Physical Attacks for Water Distribution Systems

Abstract

Modern technologies adopt Internet of Things (IoT) devices to increase water management efficiency and enhance water quality services. However, the limitations of IoT devices, such as small sizes and poor security, weaken the Water Distribution System (WDS) security, and many attackers compromise the critical components of WDS. Cyber-physical attacks (CPAs) are considered one of the biggest challenges that decrease the security factors in WDS by disrupting normal operations and tampering with the critical data of the water system. For instance, an attacker can change the water pump's speed, disrupting the service. An attacker can also alter the data of water quality parameters to contaminate the water. It is important to propose solutions to increase security in the WDS and defend against CPAs and security threats. Although several intrusion detection methods were proposed in the literature to detect WDS CPAs, many issues still need solutions, such as detecting attacks with smaller false alarms, minimizing the time to disclose the attacks, determining the location of the compromising components, and recovering solutions for the attacked components. Therefore, this paper proposes a model based on a deep learning algorithm called a Conditional variational Autoencoder (CVAE) to disclose CPAs and mitigate their bad effects on WDS. The proposed method consists of a neural network, an encoder to compress data, and a decoder to decompress data. The objective goal is to minimize the reconstruction error between the encoded-decoded data and the initial data. We apply the CVAE on some well-known datasets. The experiment results show that our proposed CVAE method performs better than others. After analyzing the CVAE model with other existing models, we get the highest performance by reaching %98 accuracy.