RSAGEL: A Hybrid Graph Neural Network Based Intrusion Detection Model for Software Defined Networking

RSAGEL: A Hybrid Graph Neural Network Based Intrusion Detection Model for Software Defined Networking

연구 분야: Networking

학회: 2024 9th International Conference on Information Science, Computer Technology and Transportation (ISCTT)

Software Defined Networking (SDN) has emerged as a revolutionary network architecture aimed at surpassing the constraints inherent in traditional network infrastructures. As SDN adoption increases, it brings additional complexities to the domain of network security. This study focuses on the detection of anomalous behaviors and the classification of various network attack types within the SDN framework. The hybrid model Residual-GraphSAGE-LSTM(RSAGEL) introduced in this paper provides a robust approach to address intrusion detection challenges in SDN environments. The advantages of this model are utilized to effectively process graph structured data, achieving an accuracy 4.5% higher than traditional deep learning algorithms on the InSDN dataset.