Reproduce “LEOSTP: A Spatio-Temporal Traffic Prediction Framework for LEO Satellite Networks” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Shaoyou Ao, Yong Niu, Zhu Han, Cheng Li, Bo Ai
LEOSTP, a conditional diffusion model, forecasts LEO satellite network traffic by explicitly integrating dynamic geographic semantics with historical traffic patterns. It achieved a 15.91% reduction in prediction error compared to leading Transformer baselines, reaching an NRMSE of approximately 0.06 on a large-scale simulated dataset.
Key insights
- Explicitly integrating dynamic geographic semantics, such as population and local time, as conditional priors is crucial for accurate LEO satellite traffic prediction.
- A diffusion model framework provides a robust mechanism to capture complex temporal dynamics and adapt to the non-stationary nature of LEO traffic, outperforming conventional sequence models.
- The semantic information of the coverage area changes continuously, necessitating a dynamic modeling approach for external conditions rather than static attributes.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.