Andrew Simsky, Jean-Marie Sleewaegen. C/A pseudoranges anomaly in geodetic-grade receivers. GNSS 2004, Rotterdam 16-19 May 2004 C/A code bias in high-end receivers Andrew Simsky, Jean-Marie Sleewaegen (Septentrio NV, Belgium) ABSTRACT This paper shows that multipath-mitigating C/A code tracking methods can induce meter-level PRN- dependent biases of C/A code measurements, which have adverse effect on positioning. Namely, we found that some geodetic-grade receivers produce C/A pseudoranges with a bias of up to 1.65m for PRNs 7,8,15,17,21,22 and 24, when their multipath mitigation feature is enabled. Analysis of this unexpected anomaly has revealed the reason for it: the affected PRNs have the peculiarity that their autocorrelation peaks exhibit slight deviations from the ideal triangle. When multipath mitigation is enabled, this distortion is mistaken as multipath and induces the bias. This anomaly adversely affects bias-sensitive applications, such as reference stations, or satellite C/A - P1 bias determination. It is shown by live examples that code-based DGPS or WAAS positioning may show meter-level deviations when multipath mitigation is active in the receiver. This paper will describe post-processing methods used to detect the anomaly, its mechanism, and countermeasures. The results of this research had been applied to Septentrio’s receiver PolaRx2. INTRODUCTION The starting point for this work was the case of abnormally high meter-level positional deviations for zero-baseline DGPS positioning between two dissimilar receivers described in [1]. The reason for these deviations has been found to be PRN- dependent biases of C/A code for one of the receivers, which translated into meter-level biases of the positional solution of code-based DGPS. The first step in investigating this problem was to measure the biases of C/A code for individual receivers and compare them to their expected values as tabulated by IGS. In this way it was established that only one of the receivers had anomalous biases, while another one was normal. The second step was to realize that the problem occurs only when multipath mitigation feature is enabled in the receiver with anomalous biases. Further on, when we found that this problem occurs in another receiver from a different manufacturer, and that exactly the same PRNs are affected (7,8,15,17,21,22, and 24), it has become clear that we deal with a fairly general problem: some multipath mitigating techniques are sensitive to peculiarities of individual PRN codes. Many multipath-mitigating algorithms work by sensing the deviations between the actually measured and nominal shape of the main auto-correlation peaks. Primarily, the slope of the main peak, which has a shape of a linear triangle, is of interest. The value of this slope for PRN codes 7,8,15,17,21,22, 24 is slightly different from the same value for all the other PRN codes. This difference is mistaken for multipath and results in a constant bias of estimated C/A code multipath errors for these PRNs. This paper begins with the review of normal (satellite- dependent) biases of C/A codes. Then we explain the origin of anomalous biases and their impact on positioning. Once anomalous biases are known and understood, they can be easily compensated either in the receiver firmware or in post-processing. NORMAL CODE BIASES Our review of normal code biases begins with the well-known τ gd (time group delay), a measure of the average differences between P1 and P2 code ranges, or, to be more precise, a component of this difference due to satellite hardware effects. The τ gd values are broadcast by GPS satellites as part of the navigation message and are independently monitored by IGS data processing centres. Updated τ gd values are regularly uploaded to satellites. According to the GPS ICD, both P1 and P2 ranges must be corrected for the time group delay because the broadcast satellite clock corrections are referred to the iono-free code combination. Therefore the τ gd bias applies to both P1 and P2 codes, as shown in Figure 1.