A GNSS/INS/Odometer Filter Bank Using Next Generation DFMC SBAS Augmentation for Rail and Road Users

Abstract

The use of GNSS as main navigation technique is limited in railway and road applications by the harsh signal reception conditions encountered in typical environments such as urban or natural canyons, buildings, tunnels etc. Therefore, in these application fields, sensor fusion frameworks are typically deployed to take advantage of the onboard sensors measurements and complement the GNSS during outages or to mitigate the effect of outliers. The ongoing GNSS evolution towards multi-frequency and multiple-constellation operation, followed in parallel by SBAS evolution, could bring important benefits also in these challenging scenarios due to the increased number of constellations (and satellites) and the increased resilience to multipath at higher signal transmission frequencies. However, since Dual-Frequency Multi-Constellation (DFMC) SBAS signals are not broadcasted yet, experimental campaign to assess the beneficial impact of DFMC SBAS for railway and road users cannot be performed and therefore simulated approaches must be performed. In this work, the availability of the NISTB testbed, capable of simulating the EGNOS V3 augmentation, offers the opportunity to make a quantitative assessment in this regard. In this work, simulated trajectories for both scenarios were used to produce synthetic on-board sensor data and GNSS dual-frequency observations with realistic reception conditions. Then, a Kalman filtering framework to fuse GNSS, IMU and odometer data was selected as a representative user algorithm for localization and integrity monitoring. GNSS observations were augmented both with standard DFMC SBAS messages and internal computational products of DFMC SBAS. Both augmentations were simulated by the NISTB testbed in order to assess the accuracy gap with respect the non-augmented case and the ideal case when the user algorithms uses International GNSS Service (IGS) products. Mean and standard deviation values of the position errors obtained in each trial, split along the horizontal and vertical directions, were used as error metrics for the performance assessment. The obtained results confirmed that, in both railway and road scenarios, an accuracy improvement was obtained for both DFMC SBAS augmentation approaches, especially in the vertical direction which is the direction most affected by errors in the clock corrections. The accuracy level achieved was very close to the one observed when using IGS products. This suggests that, if further investigation was performed, tighter protection levels could be potentially obtained. In particular, a filter bank tuning assigning more confidence to the augmented pseudoranges with respect to the non-augmented case should be tested. This should result in lower estimated covariances and protection levels.