Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn about ZOOMSYNTH, a groundbreaking system for network traffic super-resolution that reconstructs fine-grained packet-level traces from readily available coarse-grained network counters in this 16-minute conference presentation from NSDI '25. Discover how researchers from Tsinghua University and Zhongguancun Laboratory developed the first traffic super-resolution (TSR) system capable of achieving packet-level trace synthesis using only counter traces as input. Explore the innovative Granular Traffic Transformer (GTT) model and Composable Large Traffic Model (CLTM) architecture, where CLTM functions as a hierarchical tree of GTT models with each layer performing upscaling at specific granularities to capture traffic characteristics at different resolutions. Understand how the system leverages rule-following models to comprehend counter rules like ACLs when available, guiding the generation of fine-grained traces. Examine comprehensive evaluation results demonstrating that ZOOMSYNTH achieves synthesis quality comparable to existing solutions requiring packet-level traces as input, using only second-level counter traces. See how fine-tuned CLTM models outperform existing solutions by 27.5% in anomaly detection and 9.8% in service recognition tasks, and learn about the release of the pre-trained CLTM-1.8B model weights and source code to advance future research in network traffic analysis and synthesis.
