Deep Learning Model to Defend against Covert Channel Attacks in the SDN Networks

Conventional networks have grown more complicated, making it more difficult for system administrators to administer them. For instance, contemporary networks frequently include numerous computing devices such as mobiles, clients, and server systems. These devices are used with different technologies such as wireless, wired, and network virtualization. Vendor-based methods are used to configure network equipment like gateways, routers, and switches. The topology of the network may dynamically change as a result of assaults on network users and devices' mobility, failure of devices and links, and other factors. A potential approach to handling these issues in conventional networks appears to be SDN. The design philosophy of SDN provides the mechanism for implementing configurable networks with the decoupling of data and control planes. In software-defined networking (SDN), network switches receive packet forwarding rules from a central controller. However, the rules are frequently issued through applications from different organizations. Such types of applications provides the way for major conflicts between different applications of the data plane. There is a vast impact of safety rule violations between SDN components. Conflicts between IP networks' firewalls and applications are relatively very dangerous in the SDN environment. To enforce all rules in the data plane in the real-time, a rule conflict resolution should only cause a small amount of process delay. Applications running on the controller, some of which may be malicious or unstable, might be exploited to create new rules that clash with security rules already imposed in the SDN data plane. The identification of convert channel attacks, which are the primary cause of rule conflict in SDN, is the major focus of this study. In this study, a machine learning approach is suggested for detecting and preventing conversion channel-based assaults on the data plane in SDN systems. The application of the machine learning yields superior outcomes in the detection of attacks according to experimental findings.