Deep Learning based Attacks Detection of DNP3 Protocol | ||||
Aswan University Journal of Sciences and Technology | ||||
Volume 2, Issue 2, December 2022, Page 37-47 PDF (496.26 K) | ||||
Document Type: Original papers | ||||
DOI: 10.21608/aujst.2022.174148.1003 | ||||
View on SCiNiTO | ||||
Authors | ||||
Ahmed G. Yahia 1; Adly S. Tag Eldien2; Nasser M. B. Abdel-Rahim3 | ||||
1Communications & Electronics, Engineering, Helwan university, Cairo, Egypt | ||||
2Electronics, Faculty of Engineering, Benha University, Cairo, Egypt | ||||
3Power Electronics, Faculty of Engineering, Benha University, Cairo, Egypt | ||||
Abstract | ||||
Abstract. SCADA systems contain many important components that communicate with each other through communication protocols designed for SCADA systems. This paper concerns distributed network protocol 3 (DNP3), which is considered a sufficient, trustworthy, and standard protocol for improving communications between multiple vendors. The vulnerabilities of this protocol form a disaster threat over the whole system, so this paper mentions these weakness points of this protocol. Also, the paper mentions the different types of attacks that exploit these vulnerabilities. So, it is necessary for researchers to continuously study mitigating these attacks without affecting the efficiency of the system. This goal is introduced in deep learning model algorithms dependent on neural networks. This paper introduces an ensemble deep learning algorithm (autoencoders) with decision tree (DT) multiple classification and support vector machine (SVM) multiple classification. After that, applying these two classifications models to a dataset to study the efficiency of each model and compares the results between each of them using performance metrics of deep learning algorithms and confusion matrixes which show the accuracy of each classifier. | ||||
Keywords | ||||
DNP3; Autoencoders; Decision tree (DT) classification; Support vector machine (SVM) classification | ||||
Statistics Article View: 98 PDF Download: 256 |
||||