GBAS heavy-tail error overbounding with GARCH model

摘要

To reduce the inflation for statistical uncertainty and describe the real error distribution objectively, generalized autoregressive conditional heteroskedasticity (GARCH) model is utilized in this paper to model and overbound ground based augmentation system (GBAS) heavy-tail errors. Based on the GARCH model, heavy-tail errors are normalized to the standard Gaussian distribution, and error samples from all elevations are mixed together to calculate overbound without being grouped. By this means, compared with classic error distribution models, the heavy-tail errors are overbounded more tightly, and the calculated inflation factors, error confidence limits in pseudorange domain and protection levels in position domain are reduced correspondingly.