Although classical Global Navigation Satellite Systems (GNSS) based positioning provides fairly good performance in open sky conditions, the performance of the single-constellation approaches under the presence of multipath and non-line-of-sight (NLOS) signals is often not sufficient to satisfy the stringent requirements of safety-critical applications, such as the driver assistance functionalities
for inland vessels. Thus, Receiver Autonomous Integrity Monitoring (RAIM) was developed to assess the integrity of
GNSS signals, becoming the standard approach for the mitigation of errors in navigational applications. However, this Approach is known to deteriorate when handling multiple simultaneous faults, which is the most prominent case in challenging scenarios. As an alternative to RAIM for the provision of reliable solutions, the robust regression framework is presented. This framework serves as an approach for the mitigation of unexpectedly large errors on observations which do not fit to the assumption of Gaussianity of the observation noise. Here the multi-constellation approach provides access to more observables and the quality of the positioning solution can be maximized by discarding the observations not fitting the model. Despite a comprehensive work has been developed regarding a single-constellation approach and robust methods, the extension of that work for multi-constellation
using real data and outliers is still missing. Thus, this work provides an alternative framework for mitigation of errors to satisfy reliability, accuracy and availability of the positioning calculation in challenging environments. This work provides an in-depth discussion about the scale S estimator, an effective scheme from the statistical regression framework. The fault detection and
mitigation performance of the S estimator against other competing algorithms is shown. This evaluation is carried out using real data from a measurement campaign in the Moselle River in Koblenz (Germany), where the presence of several bridges induces severe multipath effects on the signals.