The GNSS integer ambiguities: estimation and validation
The GNSS integer ambiguities: estimation and validation
The GNSS integer ambiguities: estimation and validation
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
success rate<br />
1<br />
0.9<br />
0.8<br />
0.7<br />
0.6<br />
0.5<br />
0.4<br />
0.3<br />
0.2<br />
0.1<br />
upper bounds<br />
ADOP<br />
region<br />
simulation<br />
0.1 0.2 0.3<br />
f<br />
0.4 0.5<br />
success rate<br />
1<br />
0.9<br />
0.8<br />
0.7<br />
0.6<br />
0.5<br />
0.4<br />
0.3<br />
0.2<br />
0.1<br />
lower bounds<br />
bootstrapped<br />
region<br />
simulation<br />
0.1 0.2 0.3<br />
f<br />
0.4 0.5<br />
Figure 3.8: Upper <strong>and</strong> lower bounds for the success rate in the 2-D case as function of f<br />
with vc-matrix 1<br />
f Qâ,ref.<br />
Evaluation of the approximations<br />
In order to evaluate the lower <strong>and</strong> upper bounds of the success rate, simulations are<br />
used. <strong>The</strong> procedure is explained in appendix B. <strong>The</strong> results were also presented in<br />
Verhagen (2003).<br />
First the two-dimensional case is considered, using:<br />
Qˆz = 1<br />
<br />
0.0216<br />
f −0.0091<br />
<br />
−0.0091<br />
,<br />
0.0212<br />
0 < f ≤ 1<br />
for different values of f. <strong>The</strong> results are shown in figure 3.8. <strong>The</strong> left panel shows the<br />
two upper bounds <strong>and</strong> the success rates from the simulations. Obviously, the ADOPbased<br />
upper bound is very strict <strong>and</strong> is always much better than the upper bound based<br />
on bounding the integration region. <strong>The</strong> panels on the right show the lower bounds. It<br />
follows that for lower success rates (< 0.93) the bootstrapped success rate is the best<br />
lower bound. For higher success rates the success rate proposed by Kondo works very<br />
well <strong>and</strong> is better than the bootstrapped lower bound.<br />
Table 3.2 shows the maximum <strong>and</strong> mean differences of the approximations with the<br />
success rate from simulation. From these differences it follows that the ADOP-based<br />
approximation, the bootstrapped lower bound <strong>and</strong> the ADOP-based upper bound are<br />
best.<br />
Because of its simplicity the geometry-free model is very suitable for a first evaluation,<br />
Quality of the <strong>integer</strong> ambiguity solution 43