Bernese GPS Software Version 5.0 - Bernese GNSS Software

13.3 Determination of GNSS Code Biases
13.3.1 P1–P2 Code Biases
13.3 Determination of GNSS Code Biases
Based on Table 13.1, one may draw the conclusion that the geometry-free (L4) linear combination
is the most appropriate linear combination for accurate P1–P2 code bias retrieval.
P1–P2 DCB values are computed while solving for ionosphere parameters (see also Chapter
12). Remark: It should be clear that BC1−P2 (not BP1−P2) will result as receiver bias
in case of C1/P2 receiver models.
Note: Extraction of GPS τGD values is possible using RXNBV3. τGD values are exported as
BP1−P2 values in form of a DCB file.
13.3.2 P1–C1 Code Biases
It is obvious that differences between pairs of simultaneous P1 and C1 code measurements—
as provided by the first (P1/P2) receiver class—may be analyzed to retrieve inter-satellite
P1–C1 code bias values, BP1−C1. As a matter of fact, this is the method commonly used by
people doing P1-C1 bias retrieval (see, e.g., [Gao et al., 2001; Jefferson et al., 2001]). The
necessity for accounting in addition to satellite also for receiver bias values may be easily
understood in view of this straightforward method. Note that this method is not supported
by the Bernese GPS Software Version 5.0 .
At CODE, we use a fundamentally different, more sophisticated approach. Our method
works on the basis of the ionosphere-free (L3) linear combination in the course of a global
GNSS clock analysis. We solve for (satellite-specific) BP1−C1 parameters together with
the epoch parameters responding to satellite and receiver clock offsets [Schaer, 2000]. The
partial derivatives with respect to these satellite DCB parameters are exactly those factors
as they may be gathered from from the L3 row of Table 13.1: 0, or undefined for P1/P2, −1
for C1/X2, and −2.55 for C1/P2. Obviously, a “mixed” receiver network is indispensable
for our method.
The main advantage of our approach is that resulting P1–C1 DCB estimates directly reflect
the code bias differences between the three receiver classes as seen by an analysis center in
its clock estimation procedure. In other words, the bias values are estimated directly from
the data sets for which they will be applied. We do not make use of C1 code measurements
from P1/P2 receivers (providing C1/P1/P2).
13.3.3 Verification of the Receiver Tracking Technology
A special application of the P1–C1 bias parameter estimation is the possibility to verify
which of the three receiver classes a particular receiver model may be attributed to. The
principle is relatively simple: instead of solving for BP1−C1 parameters, these parameters
are assumed to be known (see also Figure 13.4) and one common factor is set up as unknown
parameter (called “P1–C1 DCB multiplier”) for each single receiver.
Figure 13.5 shows an excerpt of a corresponding GPSEST output file. A factor close to 0
indicates a P1/P2 receiver, a factor around 1 a C1/X2 receiver, and, with an expected factor
of 2.55, identification of C1/P2 is utmost reliable [Schaer, 2002].
Bernese GPS Software Version 5.0 Page 285