Abstract:Most nuclear power plants (NPPs) are looking deploying digital instrumentation and control (I&C) systems, which allow for more precise control and more economical operation. However, both the quantity and capability of industrial control system (ICS)-targeted cyber-attacks have grown dramatically over recent years. Therefore, one of the most significant challenges that digital I&C systems bring is the issue of cybersecurity, which should be enhanced before their deployment.Several different types of cyber-attacks can be introduced to NPPs; a false data injection attack on key equipment is the focus of this research due to the potential severe consequences associated with such an attack. In false data injection, the attackers may alter the reading of control sensors or commands to change the operation of an NPP. Current cybersecurity efforts focus on intrusion prevention by firewalls or data-flow direction control and use commercial intrusion detection systems, which usually focus on monitoring Internet Protocol (IP) addresses, ports, and payload length. However, attention should be given to conditions where these approaches can fail, such as an insider attack. Previous research based on process data shows the potential of a last defenseâ, line using online monitoring of the process data in concern with cyber data analysis. However, existing models involve different subsystems across the whole NPP, which has a wide attack surface and may require high computing cost. This holistic approach may not meet the time-sensitive requirements imposed upon I&C systems. This paper proposes a localized kit for key equipment in a process as a complementary detection method to improve the robustness of key equipment under cyber-attacks. Compared to existing models, this reduces the number of variables used in the model and significantly improves the computational speed. It also reduces the attack surface by limiting the data acquisition locally. This localized kit includes a cyber-attack detection model to detect anomalies within key components, such as the control system actuator, and an inference model to potentially reconstruct a compromised signal to allow the safe shut down.To develop and demonstrate the localized cybersecurity kit, a hardware-in-the-loop (HIL) testbed was built with a pressurized water reactor (PWR) simulator and a programmable logical controller (PLC). The PLC was programmed to control the steam generator (SG) water level at a specified set point, and the PWR simulator was utilized to simulate the nuclear system and response for parameters outside of the SG. Three false data injection attacks were conducted towards the testbed to generate the data needed for the localized kit development and evaluation. The results show the cyber-attack detection model is effective under false data injection scenarios and the inference model is promising as a signal reconstruction method.

Anomaly detection for early ransomware and spyware warning in nuclear power plant systems based on FusionGuard

Adversarial Attacks and Mitigation for Anomaly Detectors of Cyber-Physical Systems

Anomaly Detection and Analysis in Nuclear Power Plants

Anomoly Detection for PLC Based on Magnetic Side Channel

Fusing Design and Machine Learning for Anomaly Detection in Water Treatment Plants

NIPAD: a Non-Invasive Power-Based Anomaly Detection Scheme for Programmable Logic Controllers

Encryption-Aware Anomaly Detection in Power Grid Communication Networks

GuardFS: a File System for Integrated Detection and Mitigation of Linux-based Ransomware

A Comparative Study of Anomaly Detection Techniques for IoT Security Using Adaptive Machine Learning for IoT Threats

Machine Learning Assisted Bad Data Detection for High-throughput Substation Communication

Anomaly Detection for Insider Attacks From Untrusted Intelligent Electronic Devices in Substation Automation Systems

Robust localized cyber-attack detection for key equipment in nuclear power plants

Wavy-attention network for real-time cyber-attack detection in a small modular pressurized water reactor digital control system

Advanced Machine Learning-Driven Security and Anomaly Identification in Inverter-Based Cyber-Physical Microgrids

A Deep Spiking Neural Network Anomaly Detection Method

Anomaly Detection for Industrial Control Networks using Machine Learning with the help from the Inter-Arrival Curves

Python TensorFlow Big Data Analysis for the Security of Korean Nuclear Power Plants

Enhancing Critical Infrastructure Security: Unsupervised Learning Approaches for Anomaly Detection

Inter-Domain Fusion for Enhanced Intrusion Detection in Power Systems: An Evidence Theoretic and Meta-Heuristic Approach

A watchdog model for physics-based anomaly detection in digital substations

SPIDARman: System-Level Physics-Informed Detection of Anomalies in Reactor Collected Data Considering Human Errors