Safety Analysis of Ranging Biases on the WAAS GEOs

Abstract

The Wide Area Augmentation System (WAAS) [1] has found that the ranging signals from its geostationary (GEO) satellites can significantly improve the availability of vertical guidance, particularly in Alaska and at times when not all GPS satellites are operational. However, WAAS has also observed that the GEO ranging sources can be affected by errors that are bias-like in their behavior [2] [3] [4] [5] [6] [7]. Such errors do not change values randomly but may persist with a particular sign and magnitude for many hours or longer. Some of these bias errors commonly affect our reference receivers and may thus be difficult to observe and bound in real time. Others are readily observable but not necessarily easy to eliminate as they may impact user receivers differently. One such error results from incoherence between the code and the carrier signals. Unlike GPS, the GEO signals are generated on the ground and have to traverse the ionosphere both on the way up from the ground uplink station (GUS) to the GEO and then on the way down from the GEO to the user. The GUS electronics may not always be able to keep the two components perfectly aligned. This results in a code-carrier incoherency (CCI) that creates a varying error for users with different smoothing times. A user whose carrier smoothing filter has converged will see a different effect from a user who has not smoothed their code measurements with carrier data [2]. When WAAS generates a confidence bound on the ranging accuracy of the GEO satellites, it must account for all different users and for every error source. Unfortunately, the protection level equations used by WAAS do not support the inclusion of bias terms or terms to account for different smoothing times [8]. Therefore, WAAS must conduct special analyses to bound these biases. This paper describes the analysis WAAS performs to ensure that the UDRE it broadcasts for each GEO safely bounds all users for all possible bias errors. This analysis accounts for other fault modes that may also be present, but not yet detected by the WAAS integrity monitors. Versions of GEO bias analyses have existed since before WAAS was commissioned in 2003. The analysis has been updated and significantly improved since those early more conservative approaches. WAAS is in the midst of replacing all three of its GEOs and will briefly have four operational ranging GEOs in the summer of 2019. Pseudorange bias terms can lead to much bigger user position errors when there are more such terms that may all align. This WAAS GEO bias analysis has been recently updated and each new GEO has been carefully examined to ensure the continued safe operation of GEO ranging. This paper describes this analysis and demonstrates the safety and performance of the new WAAS GEOs Error Bounding Analysis Because the broadcast sigmas (User Differential Range Error (UDRE) and Grid Ionospheric Vertical Error (GIVE)) are larger than the actual overbounds, constant biases up to a certain magnitude can be tolerated by the user. The analysis in this paper seeks to find the maximum tolerable biases that can be present for arbitrary geometries. The analysis in this paper examines seven cases listed in Table 1. All seven cases must pass and each case itself has numerous subcases testing different UDREI and GIVEI values. Further the risk is evaluated against three user protection level calculations: the vertical and horizontal precision approach (PA) protection levels and the non-precision approach (NPA) horizontal protection level. Table 1: GEO Bias Threat Cases from Case Number Description Ionospheric State Faulted SV 1 Fault-free Performance Both Nominal and Storm None 2 CCC Fault Nominal-only GPS 3 CCC Fault Nominal-only GEO 4 SQM Fault Nominal-only GPS 5 SQM Fault Nominal-only GEO 6 UDRE Fault Nominal-only GPS 7 UDRE Fault Nominal-only GEO The fault cases evaluate whether faults will be safely detected even in the presence of the unfaulted GEO biases. Probability of Position Error Exceeding the Protection Level The purpose of the GEO Bias gaussian bounding analysis is to ensure that the broadcast UDRE and GIVE values are at least as large enough to bound the user’s actual errors. This can be written mathematically as: