Malicious Host Detection in Software Defined Networks Using Machine Learning Algorithms

Abstract

The automation and flexibility provided by the Software Defined Networks (SDN) model are crucial for addressing the challenges currently associated with the network operation. SDN network operates separately with control planes in which the control programs are used to forward the network packets from the data plane, where forwarding hardware only executes these decisions. The control programs are closely associated with each device dispersed across the network prior to the invention of SDN. SDN is being used in a large number of real-world applications as a promising network virtualization technology. The concentration of network control operations provides the features of network programmability, flexible network management, agility, and encourages network innovations are the main innovation of SDN. Software defined networks mainly focuses on providing networking functionalities by utilizing a centralized SDN controller that interacts with programmable network components. SDN made networks easier to use and more convenient, but it also invented different security flaws. These flaws could result in security risks that would be disastrous for the architecture of the SDN in particular and the entire network in general. For future networks, SDN security is essential. A reliable real-time approach is required to detect infected SDN devices. The number of unauthorized requests made to the controller can be monitored and limited using a few known techniques. These countermeasures, however, are unable to effectively defend against a series of attacks that each employs a different IP address. In order to identify malicious devices in the SDN network and to prohibit hosts and services in the SDN environment, this research presented a machine-learning approach. The results of the experiments demonstrate how well the suggested strategy classifies harmful devices.