08.06.2013 Views

Bernese GPS Software Version 5.0 - Bernese GNSS Software

Bernese GPS Software Version 5.0 - Bernese GNSS Software

Bernese GPS Software Version 5.0 - Bernese GNSS Software

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

14. Clock Estimation<br />

τ (BRUS-PTBB)<br />

100 s 1000 s 10000 s<br />

Interval<br />

Legend:<br />

Observation types used<br />

original code only<br />

smoothed code only<br />

smoothed code and phase<br />

phase only<br />

Figure 14.1: Allan variance for the time transfer between PTBB and BRUS for day 03-<br />

1390 of the example campaign using original code, smoothed code, and phase<br />

observations for the clock estimation.<br />

Precise Time and Frequency Transfer Using Phase Measurements<br />

The phase observations are much more accurate than the code data. Therefore, it is preferable<br />

to use them also for the estimation of clock parameters. The problem of using the phase<br />

data for time transfer is a one-to-one correlation between the clock parameters (cδk and cδi )<br />

and the initial phase ambiguity λn i k<br />

which is evident from the observation equations (2.34).<br />

This correlation prevents the direct access to the clock parameters if only carrier phase<br />

measurements are used. As a consequence the initial phase ambiguity parameter n i k absorbs<br />

the mean reading of the clock averaged over the measurement resp. analysis time interval. It<br />

is the change of the clock values with respect to a reference epoch which can be estimated<br />

from carrier phase observations only because the initial phase ambiguity cancels out by<br />

differencing measurements from successive epochs. This means that carrier phase alone can<br />

be used for frequency transfer only as long as phase ambiguities are connecting the epochs.<br />

On the other hand the pseudorange measurements have a direct access to the clock parameters.<br />

It is, therefore, possible to use both observation types in a combined data analysis. The<br />

different accuracy level of the two measurement types are taken into account by weighting<br />

the data in the parameter estimation procedure.<br />

The same fact may also be explained in another way: In the observation equations the<br />

correction of the receiver (and satellite) clocks with respect to <strong>GPS</strong> system time δi (δ k )<br />

contain not only the difference of the clock to the <strong>GPS</strong> time but also the hardware delays 1 .<br />

These delays may be different for pseudorange and phase observations. In the case of time<br />

transfer – i.e., when comparing two receiver clocks – the hardware delays for a satellite<br />

cancel out if it was observed from both stations at the same epoch.<br />

In a time transfer solution using only the carrier phase observations the hardware delays will<br />

be absorbed by the initial phase ambiguities. For a pure pseudorange solution, on the other<br />

hand, the hardware delays remain in the clock parameters. A combined solution using phase<br />

and pseudorange observations is necessary (a) to compute the receiver clock parameters and<br />

(b) to take advantage of the high accuracy of the carrier phase observations.<br />

1 “Hardware delay” is used in the sense of the constant part of the clock parameters in the <strong>GPS</strong> observation<br />

equations. It contains not only the real hardware delay – e.g., cable delays – but also a constant reading of<br />

the receiver (satellite) clock. The two cannot be distinguished by analyzing <strong>GPS</strong> data without changing<br />

the hardware configuration and additional measurements.<br />

Page 290 AIUB

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!