Comparative Analysis of 1D-CNN and 2D-CNN for Network Intrusion Detection in Software Defined Networks

A network intrusion detection system (NIDS) is crucial for computer networks by identifying and protecting against malicious activities and abnormal attacks. Recently, there have been several attempts to utilise convolutional neural networks (CNNs) for NIDS in Software Defined Networks (SDNs). However, these have not investigated the performance of 1D-CNN and 2D-CNN in terms of parameter number and complexity time for the same structure of layers. As a result, any comparisons may well lead to incompatible and inaccurate results. In this paper, we investigate the use of CNNs for NIDS and compare various performance parameters and time complexity for intrusion detection events. The results show that a 1D-CNN achieves 99.32% classification accuracy, surpassing the 2D-CNN and other evaluated methods. On the other hand, 2D-CNN offers less computation time when compared with 1D-CNN indicating better performance.