Effects of Preliminary DFMC Multipath Models on ARAIM Performance

Abstract

A current baseline of ARAIM from the WG-C Advanced RAIM Technical Subgroup (ATSG), described in the recently updated (v3.1) Reference Airborne Algorithm Description Document [1] makes use of the Airborne Accuracy Designator Model B (AAD-B) [2] in order to characterise the multipath contribution and receiver noise term for GPS L1/L5 and a lookup table of multipath and noise error for Galileo E1b and E5a signals, dependent on satellite elevation. The AAD-B model is derived from GPS L1 measurements only and does not characterise the effects of noise or multipath on the modernised signals, those being Galileo E1b, E5a and GPS L5. Therefore the use of the AAD-B model requires updating to ensure it is representative of the nominal errors experienced on these other measurements. Initial dual-frequency, dual-constellation (DFDC) multipath models have been presented in [4] and [5] using measurements from simulations and GNSS receivers including DFDC avionics hardware. Work is currently being conducted within the European Commission funded Dual Frequency Multipath Model for Aviation (DUFMAN) project to produce new multipath models that characterise the performance of each of the aforementioned signals, based on the collection and processing of data from aviation environments using aviation GNSS receivers. The purpose of these new models is to provide GNSS aided aviation navigation algorithms such as ARAIM and DFDC Satellite-Based Augmentation System (SBAS), with a complete DFDC multipath model to safely overbound integrity risks emanating from multipath for GPS L1c/a, L5, Galileo E1b and E5a signals. This paper shall present the findings of an investigation of the impact of these preliminary DFDC multipath models from the DUFMAN project on the performance of ARAIM by simulating global availability against the most stringent RNAV approach requirements; Localiser Performance with Vertical guidance (LPV)-200. In addition, we will present a series of comparisons to a reference control simulation when using not only the new multipath models but also alternative methods of modelling user antenna bias terms (a component of the analysis performed within the frame of DUFMAN) and finally arriving at a consideration of user antenna bias as deterministic and predictable to a degree of certainty.