GBAS protection level calculation with GARCH model

Abstract

To eliminate the time correlation and model the heavy distribution tail of ground based augmentation system (GBAS) errors, a method utilizing generalized autoregressive conditional heteroscedasticity (GARCH) model is introduced in this paper. Considering the statistical uncertainty of model parameters, a strategy for using the GARCH model in nonstationary situations is proposed. Based on that, a protection level calculation framework is established with an online/offline structure to calculate error overbound and protection level in real time. As the heavy-tail errors are normalized to standard Gaussian distribution, and all the normalized errors from different satellites and elevation groups are mixed together to calculate Gaussian overbound, the Gaussian overbound is much tighter than the one calculated by classic heavy-tail errors. That leads to smaller protection levels and higher system availability.