A Survey of AI-Based Attack Detection Models on the Edge-IIoTset Dataset

The growth of the Industrial Internet of Things (IIoT) brings new cybersecurity challenges. IIoT systems, which connect a vast number of devices, are vulnerable to cyberattacks that can lead to system malfunctions, data breaches, and privacy violations. Traditional Intrusion Detection Systems (IDS) may struggle to detect attacks in IIoT due to the complexity arising from diverse devices, protocols, and evolving attack methods. This paper focuses on studying recent research on using Artificial Intelligence (AI) for anomaly detection in IIoT systems. We leverage Edge-IIoTset, a state-of-the-art dataset specifically designed for IIoT applications. The analysis reveals the critical role of feature selection in model performance. Studies have shown that feature selection techniques can achieve high accuracy compared to models using a larger number of manually selected features. Combining selection techniques further reduces the number of features while improving performance. Privacy preservation is vital, especially in IIoT systems. Federated Learning (FL) with Local Differential Privacy (LDP) can help achieve this goal.