High-Precision Satellite Position Interpolation: A Comparison of Lagrange, Cubic Spline, and Artificial Neural Network Methods

For advanced Global Navigation Satellite System (GNSS) applications like determining the bending angle in GNSS radio occultation, it is imperative to obtain high-precision satellite position estimates at millisecond intervals. This study evaluates Lagrange interpolation, cubic spline interpolation, and Artificial Neural Networks (ANN) to achieve such high precision. Although both cubic spline interpolation and ANN are effective, ANN outperforms cubic spline interpolation with lower maximum errors and better handling of boundary conditions. In comparison, Lagrange interpolation demonstrates a lack of stability. ANN have been identified as the most effective technique for the accurate and continuous estimation of satellite positions through interpolation.