AI‑Driven Forecasting of GNSS Satellite Clock and Ephemeris Errors

The project aims to design, implement, and validate generative AI/ML models that predict the time‑varying discrepancy between broadcast (uploaded) and modeled (ICD) values of satellite clock bias and ephemeris parameters for GNSS constellations. Using a seven‑day historical error dataset, the solution will forecast errors at 15‑minute intervals up to 24 hours ahead for an unseen eighth day, improving positioning accuracy for navigation‑critical applications.