StarRTT predicts Starlink RTT in real time without ID blind zones
Original title: StarRTT: An End-to-End Real-Time Framework for Blind Zone-Free RTT Prediction in Starlink LEO Satellite Networks
RT-LSTM reports 1.405 ms mean absolute error, cutting error versus LSTM, StarNet, and ASTM.
Starlink-style LEO networks hand over often, so RTT prediction needs continuous serving-satellite data. StarRTT identifies the serving satellite every second and feeds RT-LSTM historical RTT plus TLE-propagated future satellite and gateway features. In real-time back-testing, mean absolute error is 1.405 ms over 15 horizons and 15 time-slot indices, 4.3%, 11.6%, and 22.1% below LSTM, StarNet, and ASTM.
- The gain over LSTM is 12.5% at the scheduling instant.
- Prior methods finish satellite identification only at the end of a cycle.
- A historical LSTM and a future LSTM split past and TLE features.
- Results are averaged over 15 prediction horizons and 15 reference slots.
From the original article
1 min read at arxiv.orgLow Earth Orbit (LEO) satellite mega-constellations such as Starlink require frequent handovers due to high orbital velocities, resulting in substantial Round-Trip Time (RTT) fluctuations. Accurate RTT prediction can mitigate these fluctuations, yet it fundamentally relies on continuously updated serving- satellite information.
By Jaeyong Lee · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.