Range Biases on the WAAS Geostationary Satellites

Due to limited receiver bandwidth and tracking configuration differences, measurements made on a narrowband geostationary satellite signal may differ significantly from those made on GPS signals and result in relatively large range bias errors for users. Modeling and analysis of this net effect is complicated by several parameters including a wide range of allowed GEO signal correlator spacings (relative to GPS for both the reference and user receivers). Further, the differential group delays of the reference receivers, user receivers, and GEO signal itself play a significant role in the magnitude of this effect. The result is that a narrowband GEO signal appears distorted relative to wideband GPS signals. This paper describes the history and identification of the GEO bias problem in WAAS and its current remedy. The paper models several filters and identifies the differential group delay as the key cause of this distortion. It then analyzes the effects of differential group delay on correlation functions and provides supporting analysis using actual filter models and live signal data processed in real GPS receivers. Finally, this paper provides estimates on ranging performance for future (wideband) GEOs and offers recommendations on receiver configurations designed to minimize this bias error. INTRODUCTION The Wide Area Augmentation System (WAAS) was declared operational in July of 2003. It uses a network of 25 reference stations at known locations to correct for errors in GPS satellite ranging. It then broadcasts the corrections to WAAS users via two Inmarsat geostationary satellites (GEOs). In addition to using the GEOs as datalinks, WAAS also uses them as additional ranging sources. Despite this dual intention, the GEOs have proven less accurate than GPS satellites as ranging sources. There are several differences between GPS and GEO signals. First, the civilian GPS signal is approximately 20MHz (or more) wide while the current GEO signals are only 2.2MHz. Second, the GEO signals are at relatively low elevation angles (between 9° and 30°) to most users in CONUS and are effectively stationary. Both of these factors make multipath mitigation significantly more challenging for GEO signals. Biases noted on GEO signals are often attributed to multipath. The “unobservable bias,” measured on GEO range residuals from receiver to receiver (at WAAS reference stations) was attributed to standing-wave multipath. As a result, WAAS currently adds a penalty of 5m to every GEO range measurement. This penalty increases the GEO UDRE and reduces availability for certain geometries where an additional satellite would be most useful. Further investigation of this WAAS “unobservable bias” along with Signal Quality Monitoring (SQM) research has instructed that differences in the structure of the GEO signal relative to GPS must also cause a bias for some users[2]. Any additional ranging source not identical to the GPS signals will appear as a deformed signal and can potentially create a bias in the receiver. The size of the bias will depend on the specifics of a receiver’s precorrelation filter and tracking loops. This bias will not be a common mode; it will not be removed by (i.e., included with) the clock term in the navigation solution. BACKGROUND Figure 1 shows the power spectra for both a GPS signal (PRN03) and a GEO signal. They are appreciably different, and they cause different distortions of correlation peaks as a result of the filtering differences in their respective transmission paths. As a result, both user and reference receivers are susceptible to biases that differ as a function of their receiver configurations. 156