Towards a Digital Twin for LEO Non Terrestrial Network Performance

The rapid expansion and increasing complexity of Low Earth Orbit (LEO) satellite constellations demands new digital twins for accurate predictive network performance evaluation. Modern LEO networks feature highly dynamic topologies, dense satellite deployments, and integration with terrestrial 5G/6G systems — that well-exceed current open-source/broadly available simulation tool capabilities. A high-fidelity Digital Twin model should capture real-time mobility considerations for time-varying LEO-to-earth downlink, emerging regulatory considerations (e.g. spectrum allocation driven coexistence/interference), orbit-aware inter-satellite coordination aspects, to feed into future AI-driven resource optimization architectural solutions. This talk will start by pointing out deficiencies in current 3GPP TR 38.111 and 6G-NTN EU Doppler spread models and its consequent impact downstream on NTN Downlink PHY Layer 1 processing and highlight ongoing efforts via ns-3-leo full stack network simulator at UW Fundamentals of Networking Lab (https://wp.ece.uw.edu/funlab/).