Unveiling SDN Controller Identity through Timing Side Channel

• Main Authors: Kyung, Sukwha, Baek, Jaejong, Ahn, Gail-Joon
• Format: Conference Proceeding
• Language: English
• Subjects: Accuracy; Artificial neural networks; Centralized control; Cloud computing; Data centers; Fingerprint recognition; Noise; Prevention and mitigation; Software defined networking; Timing
• ISSN: 2833-0072
• DOI: 10.1109/NoF62948.2024.10741434

Software-defined networking (SDN) has revolutionized the landscape of network management by decoupling control and data planes and becoming the backbone of many IT infrastructures including data centers, cloud computing, and enterprise networks. At the same time, however, the control plane has become a prime target for adversaries due to its critical role in network operations and centralized control functions. In this paper, we demonstrate how to discover the identity of different SDN controllers, which could be leveraged for more sophisticated attacks by adversaries. Our approach adopts a timing-based side channel and deep neural networks (DNN). To achieve this, we analyze real-world SDN traffic in a research computing center and accurately identify the controllers, minimizing the impact of random noise. Despite various factors that influence controller behaviors, our fingerprinting approach achieves an average accuracy of more than 90%. Lastly, the mitigation strategies are also discussed.