Convolution Neural Network-Based Detection in Software Defined Networks

Abstract

In recent years, software-defined networking (SDN) has evolved into adaptive emerging network technology. It is both dynamic and manageable. The SDN architecture creates a distinction between the two planes, the control and data plane. The routing and processing resolution of data packets is handled by the control plane whereas the direction to the data packet and its processing is carried out by the data plane based on the conclusions of the results of the control plane in the SDN. The protocol based on the flowing property is known as OpenFlow (OF). This is responsible for the transmission between the control plane in the SDN and its switches. This is a highly competitive field and many open-source software makers are presenting us with a variety of controllers for SDN. Some examples of the same are ONOS, Ryu, and Open daylight. A paradigm shift to using deep learning techniques is seen in the computer world with such techniques being used for extensive research in various fields. Deep learning has given breakthrough results in speech and image recognition. Its use for intrusion detection is gradually starting to increase. Deep learning can be the benchmark for the next generation of intrusion detection by early and more accurate detection of zero-day attacks. Hence, we offer the DL-IDPS (deep learning-enabled intrusion detection and prevention system) over SDN networks in this study using the Convolution Neural Network (CNN). The proposed DL-IDPS is expected to detect the different kinds of attacks possible on any SDN network.