Machine Learning-Based DDoS Attack Detection on SDN-Based SCADA Systems

Authors : Esra Söğüt, Adem Tekerek, O Ayhan Erdem

Pages : 596-611

View : 111 | Download : 115

Publication Date : 2024-01-01

Article Type : Research Paper

Abstract :Supervisory Control and Data Acquisition (SCADA) systems are used to monitor and control processes in critical infrastructures. SCADA systems do not have adequate detection and defense mechanisms against developing cyber attacks and contains many security vulnerabilities. The use of SCADA systems in critical infrastructures of national and international importance means new targets for malicious attackers. In addition, the use of SCADA systems with new technologies brings new perspectives to the security world. When technologies such as SDN are integrated with SCADA systems, it brings advantages to the system in terms of manageability and programmability. However security problems also occur against attacks such as DDoS. For these reasons, it is imperative to ensure the cyber security of SCADA systems. In this study, the case of SDN-based SCADA systems exposed to DDoS attacks is discussed. Logistic Regression, K-Nearest Neighbors, Random Forest, and Support Vector Machine classification algorithms have been used for attack detection. A ready-made dataset has been studied, and accordingly, the model that makes the most accurate determination has been proposed in our study. The results show that the proposed SVM classifier model (97.2% accuracy rate) effectively detects DDoS attacks against SDN-based SCADA systems.
Keywords : SCADA, SDN, DDoS, Makine Öğrenmesi, Modbus Protokolü

ORIGINAL ARTICLE URL