Intrusion Detection System for Cloud Based Software-Defined Networks
Omar Jamal Ibrahim, Wesam S. Bhaya
Abstract
Abstract Software-Defined Networks is a programmable network architecture for the cloud programmable control plane, decoupled from its data plane, offers new possibilities for creative security measures for overall visibility of the status network. This paper leverages these capabilities of SDN and presents the software-enabled Intrusion Detection System (IDS) architecture using the consideration of SDN. It combines the advantages of machine learning with IDS to ensure a high detection rate and protect the network from attacks. The Python script was utilized the Mininet emulator to create a virtual network. Also, it has been used as an Open Daylight software as an SDN controller hosted at a Google cloud. The proposed IDS uses a Grid Search technique with Support Vector Machine (SVM) to detect anomaly of attack. The proposed work was trained on UNSW-NB15 and NSL-KDD datasets. The results show that the proposed system offers a high detection rate. With the proposed machine learning model, the detection rate becomes more than 99.8 percent of accuracy. The results show positive progress in detecting almost all possible network attacks in the SDN-based cloud environment.