jacob5412 / SCADA-Anomaly-Detection

Anomaly Detection for SCADA systems

Anomaly Detection for SCADA systems

  __________________     _____  ________      _____   
 /   _____/\_   ___ \   /  _  \ \______ \    /  _  \  
 \_____  \ /    \  \/  /  /_\  \ |    |  \  /  /_\  \ 
 /        \\     \____/    |    \|    `   \/    |    \
/_______  / \______  /\____|__  /_______  /\____|__  /
        \/         \/         \/        \/         \/ 

Paper: Varkey, M., John, J., & Umadevi, K. S. (2022, June). Automated Anomaly Detection Tool for Industrial Control System. In 2022 IEEE Conference on Dependable and Secure Computing (DSC) (pp. 1-6). IEEE. (Link).

Introduction

The security of critical infrastructures is decreasing due to the apparition of new cyber threats against Supervisory Control and Data Acquisition (SCADA) systems. The evolution they have experienced; the use of standard hardware and software components or the increase of interconnected devices in order to reduce costs and improve efficiency, have contributed to this. SCADA systems have a number of peculiarities that make anomaly detection perform better than in traditional information and communications technology (ICT) networks. SCADA communications are deterministic, and their operation model is often cyclical. Based on this premise, modeling normal behavior by mining specific features gets feasible. (Source)

Overview

File/Folder	Description
EDA.ipynb	Exploratory Data Analysis
TSA.ipynb	Time Series Anomaly Detection
data	Data Folder
output	CSV containing IQR lower and upper bound
plots	Visualizations for each individual metric and their outliers

Requirements

• Can be found in requirements file.
• Python 3.7+
• Main packages:

catboost==0.23.1
jupyter==1.0.0
matplotlib==3.2.1
numpy==1.18.4
pandas==1.0.3
plotly==2.7.0
sklearn==0.0
seaborn==0.10.1
xgboost==1.1.0
xlrd==1.2.0

Acknowledgement

The following personnel were responsible for the dataset collection:

1. SWaT – Sridhar Adepu, Kaung Myat Aung, Desmond Wan, Beebi Siti Salimah Binte Liyakkathali
2. WADI – Venkata Reddy
3. S317 – Nils Tippenhauer, Hamid Reza Ghaeini
4. EPIC – Ding Liqun, Kandasamy Nandha Kumar, Chuadhry Mujeeb Ahmed
5. BATADAL – Riccardo Taormina
6. Blaq_0 – Francisco Furtado, Lauren Goh, Jonathan Heng
7. CISS 2019 – Desmond Wan, Francisco Furtado
8. IoT – Yan Lin Aung