SBAS Error Modelling for Category I Autoland

Several Satellite Based Augmentation Systems are currently being used by the aviation community or under development around the world. WAAS which covers US airspace currently provides the capability to perform LPV operations with 200ft minima, equivalent to Cat I performance as defined by ICAO. Airbus continues to equip new aircraft with SBAS receivers to provide the capability of LPV200 without ground-based navaids. SBAS receivers are required to meet standards defined in the ICAO SARPs and RTCA DO229D and such receivers have been shown in flight trials to have the potential to autoland under CAT I conditions as certified using GBAS. Autoland is attractive to crews to alleviate the issues associated with flight crew fatigue and unfavourable operational conditions. In the frame of GBAS CAT III R&D, a Navigation System Error model has been developed by the GNSS community which has been endorsed by AWOHWG (All Weather Operations Harmonization Working Group). This NSE model is a necessary input to the GBAS enabled CAT I autoland certification. The demonstration of Autoland performance for CAT I operations using GBAS was based on Monte Carlo simulation and flight tests. The simulation must account for both the average risk with all parameters (wind, weight, runway conditions, NSE) varying nominal in a statistical sense and the limit case risk which sets one such parameter to its limit worst case value. These simulations provide results in terms of the aircraft parameters whose regulations [CS AWO 131] will then determine if the intended operations are airworthy. This paper presents the initial work in developing the GBAS approach for an SBAS enabled CAT I autoland capability. A model of the SBAS NSE is required in the nominal and limit case, as well as in the malfunction case, to be used as an input to autoland simulations. The NSE is a function of the user-satellite geometry and the corrected and smoothed pseudorange errors. Instead of the classical approach of implementing statistical variation in the pseudorange errors, the GBAS autoland methodology determines offline, through auxiliary simulation a direct model of the position domain error. The success of this approach is in reducing the number of statistical parameters varied within the autoland simulation platform. This paper presents the same methodology in the case of SBAS NSE modelling and introduces the requirements in developing an SBAS NSE model for the malfunction case.