Author

Cayden Cather

Document Type

Honors Thesis

Date of Award

Spring 2024

Abstract

In recent years, the way we interact with daily life and industrial environments has changed dramatically due to the rise of the Internet of Things (IoT), from smart homes to industrial control systems. In particular, industry devices, sensors, and actuators can be connected to form various Industrial Internet of Things (IIoT) networks and controlled using operational technologies, such as ICS and SCADA. However, the vulnerabilities and limited security features pose significant challenges in preserving IIoT systems' security as cyber-attacks increase and become more frequent, dynamic, and complex. Additionally, the lack of IIoT testbeds has made it more difficult to find efficient ways to detect cyber-attacks and assess and ensure system security. To fill the gap, this project proposes and develops an IIoT testbed with Programmable Logic Controllers (PLCs), Raspberry Pi-based devices, a water control system, a robotic arm, and open-source software for cyber security research and education. Based on the testbed, the performance of Machine learning (ML) algorithms for cyber-attack detection are investigated and evaluated. The ML algorithms are Random Forest, Decision Tree, Naive Bayes, KNN, Logistic Regression, MLP Classifier, RNN, and Transformer. Using the collected data in the testbed and the CICIoT2023 dataset, the ML modes are trained and tested to determine whether incoming packets are benign or harmful by identifying their attack type. The train-test split that best optimizes the performance of each model is also tested. This is done by comparing the accuracy, precision, F1-Score, and recall of each model based on the percentage of the data that is used for training and testing. In addition to the dataset, the CICIoT2023 dataset is used to further assess the effectiveness of ML algorithms and evaluate the models’ performance. Future work is highlighted in the conclusion to extend the testbed’s capacity and performance for IIOT cyber security study and research purposes.

Share

COinS