Link discovery attacks in software-defined networks: Topology poisoning and impact analysis

Software Defined Networking (SDN) has become a popular technology that offers advantages of programmable and flexible network management over the legacy practice. The centralized SDN controller is an important enabler of these benefits. One of the most crucial tasks of the SDN controller is link discovery as it provides topology of the network essential for the controller to direct or create rule forwarding and routing mechanisms. Much research on SDN security has been studied but only recently that security of OpenFlow link discovery protocols and topology poisoning have been addressed. Existing work includes link fabrication attacks via compromised hosts and defense systems with authentication. This paper discusses SDN link discovery process and its vulnerability to link discovery attacks including new attacks via compromised switches. We present a simple but effective defense mechanism using active ports that can detect both host-based and switch-based link discovery attacks. Finally, the paper presents an analytical and empirical analysis of the impacts of topology attacks on routing. The paper discusses attack details, proposed methods and results of these analyses.