Integrity for Future SBAS Users: Concept and Experimentations

The present paper summarizes the work carried out under the European H2020 R&D study “EGNOS Next System Feasibility study” for which a dedicated activity addresses a new integrity concept focused on the SBAS’ products in the pseudorange domain and aimed at new potential SoL users beyond the civil aviation user compliant to RTCA/EUROCAE MOPS [1], [2]. In GNSS applications, integrity is defined as a measure of the level of trust a user can place in a position estimate [3]. Th e classical Satellite Based Augmentation Systems (SBAS) aviation integrity concept relies on a stringent specification of the user algorithm implementation, described in the annex of the Minimum Operational Performance Standard (MOPS) [1], [2]. Similarly, again according to MOPS, current SBAS integrity data assumes that residual errors from broadcast corrections (satellite orbit/clock and ionosphere delays) are overbounded by a Gaussian Cumulative Distribution Function (CDF). For possible future SBAS’ applications such as maritime, rail, automotive and Unmanned Aerial Vehicle (UAV), the ability to provide tight and reliable protection volumes will be a critical enabler. To meet this level of performance, users will need to use more accurate positioning algorithms than the one specified in the MOPS for aviation users. This improved positioning service should address both the SBAS ability to broadcast highly accurate corrections and reliable integrity data, as well as the user algorithm formulation which uses these SBAS products. Moreover, these users may evolve in environments not favorable to the correct reception of GNSS signals, implying possibly a fusion with measurements coming from other sensors. In this context, the use of Kalman filtering and hybridization with non-GNSS sensors seem highly recommended. For users implementing Kalman filtering, the knowledge of the time correlation of the GNSS measurements errors is of importance to determine correct overbounding models. The current MOPS integrity concept is not adapted for such new user solutions. Typically, CDF overbounding is applicable only for GNSS measurements with residual errors not correlated with time. To cope with time correlation, recent developments introduce overbounding models of the Power Spectral Density (PSD) of the GNSS measurements residual errors[7]. In this paper we describe a new potential SBAS service which provides integrity parameters for a wide class of user algorithms. The new concept focuses on the integrity of SBAS products at pseudorange level to stay as much user agnostic as it can be, in the sense that it is demonstrated that these products are compatible with several user design solutions (variety of GNSS pre-correlation filters, variety of positioning algorithm formulations, several environments and user characteristics, etc…) and target integrity risks. For instance, these products are compatible with a Kalman filter using only code measurements and either the ionosphere -free combination or uncombined measurements, in a float ambiguity PPP Kalman filter with uncombined or ionosphere-free measurements. It combines the classical SBAS sigma parameters with PSD overbounding to enable its use by users implementing Kalman filtering. This concept consists in a combination of offline parameters and parameters estimated in real time. The offline parameters are checked by the system to confirm that the required assumptions for maintaining integrity remain verified, and can be updated by a command if needed. In addition, the SBAS navigation kernel supporting this service will implement precise orbit, clock, code, and phase bias computation for providing to the users the set of corrections compatible with Precise Point Positioning. The new potential service has been implemented in a System Test Bed that takes real measurements from IGS stations as inputs. The IGS network was chosen to mimic EGNOS and Galileo ground station networks. The integrity concept and its applicability were tested in an end-to-end experiment with real user data. In this experimentation, we consider Kalman filtering positioning using ionospheric free combination. We consider both a code only and a phase-based formulation (float ambiguity resolution). We demonstrate that the computed protection volumes are firstly reliable bounds of the position errors, and secondly compatible for more stringent applications querying smaller protection volumes.