Jul 13, 2026
Paper Presentation at IEEE NetSoft 2026
Binh presented "xChain: Multi-stage Traffic Analysis and Classification in O-RAN," introducing a novel multi-stage framework that combines lightweight traffic analysis with specialized AI models to improve both inference latency and classification accuracy. The proposed FastInfer approach achieves up to 6× lower inference latency than existing state-of-the-art methods while maintaining high classification accuracy, enabling more responsive AI-powered network control in O-RAN.
In addition to the paper, Binh demonstrated "Intelligence Where It Matters in Open RAN," an interactive platform that allows users to compare multiple AI models—including CNN, LSTM, GNN, and FastInfer—as O-RAN xApps under realistic network traffic. The live demo visualizes the trade-off between classification accuracy and inference latency in real time, helping researchers and practitioners better understand AI model selection for next-generation mobile networks.
