A Detailed Analysis of Public Industrial Control System Datasets

This is the authors' abstract. We don't add key points for this paper.

A wide range of critical infrastructures such as power systems, water distribution systems, gas pipelines, and others are controlled and monitored using industrial control systems (ICSs).Recently, security attacks against ICSs are increasing at an alarming rate.These systems cannot afford to lose the availability of service; a cyber-attack can cause catastrophic damage.Intrusion detection systems (IDSs) are the first defence line against such attacks.To develop an effective IDS, a well-designed dataset is a must.In this paper, we present a detailed analysis of public intrusion datasets for ICSs.Focusing on the way security researchers used them to develop an IDS, their results, and the effect of the dataset's drawbacks.We performed exploratory data analysis (EDA), principal component analysis (PCA), and binary classification using random forest (RF) model.We believe this analysis will help the developers of the next generation of ICS-related IDSs.