Determination of Pseudorange Error Models and Multipath Characterization under Signal-Degraded Scenarios
Kasia Gibson, German Aerospace Center (DLR), Germany/Northeastern University, USA; Daniel Medina, Ralf Ziebold, DLR, Germany; Pau Closas, Northeastern University, USA
With the need for autonomous transportation in demand, it is of great importance that the utilization of Global Navigation Satellite Systems (GNSS) provides high accuracy positioning, to ensure safe navigation. GNSS technology has been under substantial development over the last years, especially with the evolution and operation of the European Union’s Galileo and China’s BeiDou constellations. While position estimation is undoubtedly dominated by the usage of GNSS, multipath effects caused by elevation and interference factors are still a major challenge when utilizing GNSS.
Multipath has become the most pronounced source of error in terms of positioning while navigating during challenging scenarios. This is especially the case in urban canyons and other elevated regions, where the high buildings and metallic structures are responsible for multiple reflections. Moreover, given the locality of the effect, augmentation systems are unable to provide any assistance. Another concerning factor surfaces when integrity monitoring is intended. An erroneous variance of the GNSS measurements can result in the complete failure of the integrity estimation. Similarly, the integration of GNSS in a multi-sensor scheme is susceptible to both a lack of errors in the measurements and a correct estimation of its variances. Improper modelling of the measurements’ variances can reduce the accuracy of the estimation as much as a gross bias on these measurements. It is therefore required to have reliable and statistically meaningful modeling of errors.
The characterization of the uncertainty present in GNSS signals has been a recurrent topic in the GNSS community. Thus, in  the first approach towards integrity monitoring was discussed, including a profound description on satellite weighting. This work was later extended in , where multi-frequency, multi-constellation systems were taken into consideration, consequently updating the GNSS-satellite variance models. Although much of these works play a crucial role for the aviation domain, its application to other areas is not yet suggested, given the strong assumptions and the particular fault modes occurring in aviation. In , the question about how to monitor reliability and weight observations in complex scenarios was raised. For this case, data from a static scenario with degraded GNSS-signals was collected over a long time with the aim of formulating a new variance model based on signal strength. The usage of this model has lately become conventional for navigation in signal-degraded scenarios, as seen in [4,5,6]. In , the evaluation of the multipath and noise error using the combination of Code-Carrier and Divergence-Free smoothing was proposed, as a way to assess processing models for Ground Based Augmentation System.
This work presents novel nonlinear error models for the characterization of noise and multipath of GNSS signals. For this analysis, multi-frequency and multi-constellation GNSS data are collected over a long time in a open-sky static scenario as well as in a signal-degraded dynamic scenario, tracing the errors present in the observations. New variance models are derived from a regression and multidimensional fitting problem, which relates the positioning error traces with parameters such as the elevation angle and the carrier-to-noise ratio. The new error model and other state-of-the-art models are used as weighting schemes in a least-squares solution. Thus, the positioning performance, based upon the weighing schemes of these models, are compared to each other in a complex dynamic scenario under harsh multipath and NLOS effects. The evaluation is done in terms of positioning accuracy and fidelity of the estimated protection levels.