A Reliable Method for Detection of Compromised Controller in Software Defined Networks

Abstract

With the advancement in technology and increased innovations in network managing large networks become easy due to an emerging technology known as SDN software defined networking as accepted by many enterprises. By separation of control plane and data plane, SDN is abstracted by a new layer which is implemented by controller. By introducing more practical issues in SDN, today SDN controller is a challenging aspect in terms of security of controller hijacking, compromised controller, malicious controller. The controller is the main nerve of SDN which manages all the actions and functions. Software Defined Networking (SDN) allows network controller to handle and manage the network according to user choice and requirements. SDN Controller is infected by many serious and disastrous attacks and due to centralized behavior of SDN, Controller gets compromised by malwares and attacks. In this paper we introduce framework that completely detect compromised controller in SDN. Our basic idea is to detect percentage of malicious behavior present in network by extracting four open flow parameters from open flow traces and fed them into backup Controller for handling all update information. Finally the information is fed into Random Forest classifier that will test and check the controller is compromised or not. Following this idea and using Random Forest Classifier compromised controller is detected in network with performance analysis of 98% with better feasibility and efficiency than prior work to secure SDN.