Zero-Doppler Pseudorange Biases

Abstract
Accurate GNSS time transfer depends on the precise determination of pseudorange biases. The prominent source of pseudorange bias is the hardware delay in receiver elements (antenna, cable and receiver frontend), but other sources of bias do exist. This paper addresses a bias that has so far not received much attention in the GNSS community, and which specifically affects pseudoranges taken at small or zero Doppler. These biases are of particular relevance to time transfer based on geostationary satellites or to absolute receiver calibrations involving the simulation of geostationary satellites. The paper starts by showing that GNSS pseudoranges can exhibit an oscillating error pattern when the Doppler is smaller than a few Hertz. At zero Doppler, some satellites are shown to be affected by a constant nanosecond-level bias. Examples from
real-life datasets and GNSS simulations, and from different types of geodetic-grade receivers are given. The paper describes the nature of this type of bias. It is explained why GNSS receivers tend to have difficulties producing accurate pseudoranges when the Doppler is lower than a few Hertz. It is shown that the resulting bias depends on the receiver digital sampling frequency and on the satellite PRN code and chipping rate. The size of the bias is quantified for the different GNSS constellations and signals and for typical sampling frequencies used in GNSS receivers employed for time transfer. It is shown that the GPS L1 C/A signals are the most affected, and that the effect is satellite-dependent, with some satellites showing biases of up to 30 cm (1ns). It is then shown that the error can be predicted and compensated for when the relevant receiver parameters are known. The pseudorange errors from an uncorrected receiver and from a receiver applying a zero-Doppler bias compensation are compared, demonstrating the accuracy of the compensation.