Network Digital Twin
NDTwin is a novel network digital twin (NDT) open source framework designed for optimally operating and managing a network.
- Its Kernel continuously collects real-time and ultra-precision flow states using a novel zero control-plane overhead sFlow scheme (IEEE ICC 2026, to appear).
- Its Apps use simulation and AI/ML technologies to: 1) evaluate/predict the outcome of many "what-if" conditions in parallel, 2) find the optimal solution to the current or a predicted situation quickly, and 3) issue commands to network devices in real time to perform the best solution.
- Its Tools feature a Web GUI that uses Large Language Model (LLM) to support intent-based network management and a real-time network traffic visualizer.
- NDTwin operates correctly and successfully on both physical networks composed of hardware switches and emulated networks formed by Mininet.
- It can be used as an automatic system to optimize the operation of a production network or as an academic platform to conduct NDT-based research.
- Developers can use this framework to develop, test, evaluate, and deploy their innovative NDT applications.
How NDTwin Works
The Kernel
Its Kernel continuously collects real-time network, device, and flow states. It acts as the brain, ensuring synchronization between the physical network and the digital twin.
AI & Apps
Supports advanced AI/ML algorithms and LLM integration for traffic prediction, anomaly detection, and automated network optimization.
Digital Twin Powered
Employ digital twin technologies to optimize a production network.
OpenFlow Support
Compatible with OpenFlow standard. Seamless integration with Ryu SDN controller.