The GNSS integer ambiguities: estimation and validation
The GNSS integer ambiguities: estimation and validation
The GNSS integer ambiguities: estimation and validation
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Table 2.3: Availability <strong>and</strong> accuracy of GPS satellite positions, after (Neilan et al. 2000).<br />
Orbits Availability Accuracy<br />
Yuma almanacs Real-time or earlier ??<br />
Broadcast Real-time 5 m<br />
IGS Ultra-rapid Near real-time 20 cm<br />
IGS Rapid 17 hours 10 cm<br />
IGS Final 10 days 5 cm<br />
2.3 <strong>GNSS</strong> functional model<br />
In this section different <strong>GNSS</strong> functional models will be presented. Both the nonpositioning<br />
or geometry-free models <strong>and</strong> the positioning or geometry-based models are<br />
presented.<br />
Different <strong>GNSS</strong> models can also be distinguished based on the differencing that is applied.<br />
Differencing means taking the differences between observations from e.g. different<br />
receivers <strong>and</strong>/or different satellites. It is often applied in order to eliminate some of the<br />
parameters from the observation equations.<br />
2.3.1 General mathematical model<br />
<strong>The</strong> functional model describes the relationship between the observations <strong>and</strong> the unknown<br />
parameters. <strong>The</strong> m observation equations can be collected in the system y =<br />
Ax + e, where y <strong>and</strong> e are r<strong>and</strong>om, e is the discrepancy between y <strong>and</strong> Ax. It is assumed<br />
that the mean E{e} is zero, since e models the r<strong>and</strong>om nature of the variability<br />
in the measurements, <strong>and</strong> this variability will be zero ’on the average’. This gives for<br />
the expectation of y:<br />
E{y} = Ax (2.26)<br />
<strong>The</strong> probability density function of y describes the variability in the outcomes of the<br />
measurements. For normally distributed data, it is completely captured by the dispersion:<br />
D{y} = E{ee T } = Qy<br />
(2.27)<br />
This is referred to as the stochastic model, with Qy the variance-covariance (vc-) matrix<br />
of the observations. This matrix describes the precision of the observations <strong>and</strong> it<br />
is needed in order to properly weigh the observations in the adjustment process, see<br />
section 2.5. In section 2.4, the stochastic models corresponding to the functional models<br />
described in this section will be given.<br />
<strong>The</strong> functional <strong>and</strong> stochastic model of (2.26) <strong>and</strong> (2.27) together are referred to as<br />
Gauss-Markov model.<br />
14 <strong>GNSS</strong> observation model <strong>and</strong> quality control