Against the backdrop of growing demand for global ubiquitous connectivity in the 6G era, low-Earth-orbit (LEO) satellite mega-constellations have emerged as a key complement to terrestrial networks, yet their high topological dynamics and large scale pose notable challenges for accurate and efficient simulation. In a study published in Engineering, researchers from Nanjing University and the University of Waterloo introduce UltraStar, a high-fidelity and high-efficiency simulator designed to support the development and performance analysis of 6G integrated space-ground networks with LEO mega-constellations, addressing critical gaps in existing simulation tools for large-scale satellite networks.
UltraStar features a systematic, scalable three-plane architecture comprising logical, control and data planes, with an NS-3-based kernel at its core, and offers intuitive visualisation via a Web user interface. To capture the heterogeneous and dynamic topology characteristics of LEO satellite networks, the simulator integrates real ephemeris data through a custom environment update algorithm, enabling precise satellite orbit prediction using the SGP4 model and NASA's SPICE toolkit, as well as realistic modelling of sun outages, inter-satellite link (ISL) variations and ground-satellite link (GSL) handovers. It also builds a 3GPP-compliant NTN channel model for GSL and ISL link budget calculations, supporting modulation and coding schemes (MCSs) aligned with DVB and CCSDS protocols, and incorporates real data-trace-based audio and video traffic models to replicate practical application behaviours.
To overcome the inefficiency of serial discrete event simulation for large-scale networks, UltraStar adopts a message passing interface (MPI)-based parallel and distributed approach, with a dynamic look-ahead time-window parallel synchronization algorithm tailored for the high topological dynamics of LEO constellations. This design balances computational load across multiple cores or machines, avoiding event ordering and causality errors common in traditional parallel simulation methods.
Comprehensive case studies across the physical, link, network and transport layers validate UltraStar's performance. The simulator accurately models time-varying SINR for GSLs and BLER for ISLs, enables optimized handover strategy evaluation for civil aviation scenarios in space-air-ground integration, and supports the testing of enhanced routing protocols like PHBGP and transport frameworks such as the QUIC-enabled QRST, with results demonstrating precise performance assessment of key metrics including convergence delay, packet loss rate and frame completion ratio. Efficiency tests show that UltraStar supports system-level simulation of up to 2000 LEO satellites, achieving a 17.8x speedup with 24 computing cores compared to serial simulation under a near-1 Gbps traffic load, with a scale factor of no more than 60.
The researchers note that UltraStar facilitates the design and analysis of algorithms and protocols for 6G space-ground network operation and deployment. Future work will focus on applying the simulator to real-world network scenarios, enabling customized simulation and analysis to meet specific industrial and research requirements.
The paper "A High-Fidelity and High-Efficiency Simulator for 6G-Integrated Space–Ground Networks," is authored by Haibo Zhou, Xiaoyu Liu, Xin Zhang, Xiaohan Qin, Mengyang Zhang, Yuze Liu, Weihua Zhuang, Xuemin (Sherman) Shen. Full text of the open access paper: https://doi.org/10.1016/j.eng.2025.08.042
