Proc. ACM Meas. Anal. Comput. Syst. (SIGMETRICS), Vol. 10, No. 2, June 2026
Horizon is the first global-scale machine learning system for predicting LEO satellite Internet performance. By integrating crowdsourced measurements from M-Lab NDT7 and Cloudflare AIM with meteorological data and satellite orbital propagation, Horizon achieves accurate performance predictions spanning 90+ countries.
First ML system predicting Starlink performance globally using crowdsourced measurements from M-Lab NDT7 and Cloudflare AIM.
Integration of weather indices, satellite density via orbital propagation, and spatio-temporal features for comprehensive modeling.
Leave-one-location-out experiments validate predictions for entirely unseen regions with comparable accuracy.
Publicly available dataset spanning 11 months, 90+ countries, with full reproducible pipeline.
A five-phase pipeline from raw measurements to global predictions
Crowdsourced measurements, weather APIs, satellite orbital databases
Data cleaning, server filtering, cross-dataset validation
Weather Index via PLS, satellite density from orbital propagation
Location-aware filtering preserving natural variability
Ensemble combining Random Forest + Gradient Boosting
Latitude contributes 42-46% of feature importance, confirming Starlink's orbital geometry as the dominant performance factor.
The novel PLS-derived Weather Index captures 14-15% of predictive power through cloud cover, precipitation, wind speed, and temperature.
Leave-one-location-out cross-validation shows the model generalizes to entirely unseen cities with R² = 0.44 for latency.
TU Delft (now at Google)
TU Delft
TU Delft