GAST-D Monitoring Results from Post-processed Flight Trial Data - Performance Evaluation of DLR´s GBAS Testbed

Kurzfassung

The German Aerospace Center (DLR) implemented an experimental GBAS test bed with three ground reference receivers at Braunschweig-Wolfsburg research airport. Its compliancy with respect to accuracy, integrity, and availability requirements of CAT-I precision approach has been demonstrated in several flight trials in 2009. The research ground station consists of three Topcon Net G3 dual frequency GNSS receivers with multipath suppressing choke ring antennas, which deliver corrections for navigation at a rate of 2Hz. However, in the current stage only range and carrier data from the aeronautically protected L1-band is used .Stations are separated by baselines between 740 and 770 meters. Due to its high modularity, the station is fully accessible and can be easily reconfigured for research purposes. Currently, reconfiguration to an enhanced infrastructure is taking place to comply with the more stringent requirements of CAT-III / GAST-D approaches. In preparation for this reconfiguration, we post-processed the data collected during the 30 approaches of the 2009 flight trials (flown with DLR´s VFW614 aircraft) with respect to the GAST-D parameters and requirements defined in the ICAO draft SARPS and RTCA DO-253C. In addition to the CAT-I / GAST-C processing which is based on 100s carrier smoothed code pseudoranges, these requirements now include a second smoothing filter and pseudorange correction processing with 30s smoothing time on both, ground and aircraft subsystem. Moreover, a battery of new monitors on the airborne side is required to ensure a sufficient level of integrity. They include a dual solution ionospheric monitoring architecture (DSIGMA), differential correction magnitude check, bias approach monitor, reference receiver fault monitor and fault detection and exclusion. The dual solution ionospheric monitor architecture compares the position solution with 30s smoothed and corrected pseudoranges with the one obtained using the 100s Hatch filter. If the difference exceeds 2 meters, an ionosphere anomaly is considered present and the approach service is degraded to GAST-C. The differential correction magnitude check makes sure the broadcast differential corrections do not exceed 200 meters. The bias approach monitor compares the vertical projection of the 66% position uncertainty with the final approach segment vertical alert limit when entering the precision approach region. The reference receiver fault monitor once more checks the broadcast B values to identify faulty ground receivers. Fault detection and exclusion is performed by a weighted RAIM algorithm. We show the outputs of these monitors during the flight trials and evaluate their performance with the ones expected theoretically. The output of the DSIGMA monitor was analyzed and overbounded in order to obtain inflated parameters sigma_v=0.21944m and sigma_l=0.12046m, which are then used in the reference receiver fault monitor. Additionally, we assessed the ground station performance for the 30s smoothing filter. We confirm that the performance gain of a 100s filter time constant with respect to the 30s one falls much below the expected value of 1.8 since a correlated low frequency error is still present in the data. With respect to the accuracy improvement, the results suggest a re-siting of the ground station antennae to a lower height and away from metallic structures to improve ground station performance. We furthermore evaluated the integrity, availability and continuity performance during the flight trials. For this we used a dual frequency (L1-L2 semi-codeless) carrier phase positioning post-processed solution as a truth reference. No misleading information or hazardously misleading information occurred and the service was never interrupted or unavailable during any of the approaches. Position accuracy was always below 1.86m vertical and 0.91 m lateral (one standard deviation) at all times within the GBAS service area. Moreover, we have optimized the location of ground station antennae to fulfil the ICAO requirement of detecting absolute slant ionospheric gradients above 300mm/km with the double difference carrier phase monitoring architectures (as currently being discussed at the ICAO NSP panel) under the assumption of a worst case placement of the reference station 5 km away from the threshold. The optimized configuration for four receivers is linearly distributed along the runway with a maximum baseline of 220m and allows an increase of allowable carrier phase noise by 20% to a standard deviation of 8mm compared to an optimal architecture using three receivers.