Volume 14, Number 4

An Efficient Deep Learning Approach for Network Intrusion Detection System on Software Defined Network


Mhmood Radhi Hadi and Adnan Saher Mohammed, Karabuk University, Turkey


Software-defined Networking (SDN) is a new technology for changing network architecture and making it more flexible and controllable. SDN can control all tasks of a network through the controller. Providing security for controller consider extremely important. Due side of the controller on the network side Network intrusion detection system (NIDS) will be effective to provide security for the controller. In this study, we suggest building a system (NIDS-DL) to detect attacks using 5 deep learning classifiers (DNN, CNN, RNN, LSTM, GRU). Our approach depends on the binary classification of the attacks. We used the NSL-KDD dataset in our study to train our deep learning classifiers. We employed 12 features extracted from 41 features using the feature selection method. CNN classifiers harvest the highest results in most evaluation metrics. Other classifiers also achieved good results. We compared our deep learning classifiers with each other and with other related studies. Our approach achieved success in identifying the attacks and might be used with great efficiency in the future.


Network Intrusion Detection System, Software Defined Networking, Deep Learning, NIDS, SDN.