ZeekWeek 2022 - Practical GAN-based Synthetic IP Header Trace Generation w/ NetShare - Yucheng Yin
By Yucheng Yin, PhD student/CMU
We explore the feasibility of using Generative Adversarial Networks (GANs) to automatically learn generative models to generate synthetic packet- and flow header traces for network-ing tasks (e.g., telemetry, anomaly detection, provisioning). We identify key fidelity, scalability, and privacy challenges and tradeoffs in existing GAN-based approaches. By synthesizing domain-specific insights with recent advances in machine learning and privacy, we identify design choices to tackle these challenges. Building on these insights, we develop an end-to-end framework, NetShare. We evaluate NetShare on six diverse packet header traces and find that: (1) across distributional metrics and traces, it achieves 46% more accuracy than baselines, and (2) it meets users’ requirements of downstream tasks in evaluating accuracy and rank ordering of candidate approaches.
Building on top of the insights from PCAP and NetFlow, NetShare could serve as an efficient tool to share sensitive zeek logs, which could facilitate the researchers and developers to devlop more robust and accurate models with access to a broader set of data.
Speaker Bio: Yucheng Yin is a Ph.D. student at Electrical and Computer Engineering, CMU and Cylab advised by Prof. Vyas Sekar and Prof. Giulia Fanti. His research interests include the application of machine learning (especially Generative Adversarial Networks, or GANs) to networking, security, and systems. His works have appeared at several top venues like ACM SIGCOMM, USENIX SECURITY, and NDSS. Prior to joining CMU, he receives a dual bachelor’s degree from Shanghai Jiao Tong University (ECE) and the University of Michigan (CS).