Enhanced Local-Area DGNSS for Autonomous Vehicle Navigation: Optimal Smoothing Strategy

Abstract

Local Area Differential GNSS (LADGNSS) is one means of providing navigation and guidance to autonomous vehicles with very high accuracy and integrity. In previous work, the authors have developed a low-cost, portable LADGNSS system prototype based on the single-frequency (L1-only) Ground-based Augmentation System (GBAS) architecture developed for civil aviation. This work expands that system to use multiple frequencies (L5/E5 in addition to L1) and Galileo satellites in addition to GPS, creating a Dual-Frequency Dual-Constellation (DFDC) system. This creates additional options for the carrier smoothing of code phase that is critical to reducing code-phase (pseudorange) errors and provides additional means to detect and exclude signals affected by anomalous ionospheric behavior. This paper develops new models of nominal errors for DFDC LADGNSS to represent error correlation across time and among ground reference receivers. These models support detailed comparisons of different smoothing algorithms and time constants in the presence of multiple nominal error sources. Vertical Protection Levels (VPLs) for a set of candidate smoothing processes and time constants are generated for 27-satellite GPS and Galileo constellations to determine which give the best performance (lowest VPLs, and thus highest availability) under different operational scenarios, different user distances from the LADGNSS reference station, and different levels of nominal ionospheric activity (in mid-latitudes and in equatorial regions).