Abstract:
Protecting networks and systems from unauthorized access and cyber threats is increasingly critical. Intrusion Detection Systems are essential in achieving this, especially in Industrial Control Systems such as Supervisor Control and Data Acquisition to ensure the safety of Critical Infrastructures like power grids, water treatment facilities, and gas plants. This paper evaluates different Intrusion Detection System models designed for Supervisor Control and Data Acquisition protocols, including Distributed Network Protocol 3, International Electrotechnical Commission 61850, and Modbus. The objective is to provide a detailed evaluation of the current state of machine learning-based Intrusion Detection Systems and to propose a suitable model for African countries, particularly Mozambique and Senegal, where there is a need for enhanced power grid infrastructure. The study explores various Intrusion Detection Systems based on machine learning techniques such as Decision Trees, Random Forest, k-Nearest Neighbors, Support Vector Machines, and Deep Neural Networks. It analyzes the performance of these systems, discussing their strengths, limitations, and the challenges associated with them. The paper concludes that the Intrusion Detection Systems reviewed, which are based on machine learning models, showed remarkable performance. It suggests future directions to address the challenges and improve the evaluation of these models.