Detection of Cyberattacks in SCADA Water Distribution Systems Using Machine Learning: A Systematic Review of the Literature

Various industries use supervisory control and data acquisition (SCADA) systems to monitor and control different processes, one of which is water distribution systems. In recent years, intentional cyberattacks targeting these systems have increased. It is essential to protect them, and intelligent technologies, such as machine learning, can guarantee their productivity and safety. The objective of this study is to describe the different models, techniques, metrics, datasets, and machine learning algorithms applied in the detection of cyberattacks through a systematic review of the literature using the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) methodology. The databases consulted were Web of Science, Scopus, Springer, ScienceDirect, and IEEE, from which 656 articles were retrieved. An exhaustive bibliometric analysis was carried out, and the 40 most relevant articles were selected. The results show that themost used models are artificial neural networks (five mentions) and K-nearest neighbors (five mentions). In addition, one article had an accuracy metric of 99.79%.