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.
Integer Aperture <strong>estimation</strong> 5<br />
In section 3.2 the success rate was introduced. This probability is a valuable measure<br />
for deciding whether one can have enough confidence in the fixed solution. As soon<br />
as the success rate is close enough to one, a user can fix the <strong>ambiguities</strong>, so that the<br />
precise carrier phase measurements start to contribute to the solution. However, when<br />
the success rate is considered too low with respect to a given threshold, it will not yet be<br />
possible to benefit from the precise phase observations. This also means that there is a<br />
probability that the user unnecessarily sticks to the float solution just because the fail rate<br />
is too high. In practice, not only the success rate is considered to decide on acceptance<br />
or rejection of the <strong>integer</strong> solution. <strong>The</strong> <strong>integer</strong> solution is also validated using one of<br />
the methods described in section 3.5. Since no sound <strong>validation</strong> criterion exists yet, in<br />
this chapter a new class of <strong>integer</strong> estimators is defined, the class of <strong>integer</strong> aperture<br />
estimators. Within this class estimators can be defined such that the acceptance region<br />
of the fixed solution is determined by the probabilities of success <strong>and</strong> failure.<br />
<strong>The</strong> new class of <strong>integer</strong> estimators is defined in section 5.1. <strong>The</strong>n different examples<br />
of <strong>integer</strong> aperture estimators are presented. First, the ellipsoidal aperture estimator in<br />
section 5.2, for which the aperture space is built up of ellipsoidal regions. <strong>The</strong>n in section<br />
5.3 it is shown that <strong>integer</strong> <strong>estimation</strong> in combination with the ratio test, difference test,<br />
or the projector test belongs to the class of <strong>integer</strong> aperture estimators. When the<br />
aperture space is chosen as a scaled pull-in region, the <strong>integer</strong> aperture bootstrapping<br />
<strong>and</strong> least-squares estimators can be defined. This will be shown in section 5.4. Finally,<br />
the penalized <strong>integer</strong> aperture estimator <strong>and</strong> optimal <strong>integer</strong> aperture estimator are<br />
presented in sections 5.5 <strong>and</strong> 5.6. Section 5.7 deals with implementation aspects. A<br />
comparison of all the different estimators is given in section 5.8, <strong>and</strong> examples in order<br />
to show the benefits of the new approach are given in section 5.9.<br />
In order to illustrate the principle of IA <strong>estimation</strong>, throughout this chapter a twodimensional<br />
(2-D) example will be used. Simulations were carried out to generate<br />
500,000 samples of the float range <strong>and</strong> <strong>ambiguities</strong>, see appendix B, for the following<br />
vc-matrix:<br />
Q =<br />
Qˆ b<br />
Q â ˆ b<br />
Qˆ bâ<br />
Qâ<br />
Note that Qâ = Qˆz 02 01.<br />
<br />
⎛<br />
0.1800 −0.1106<br />
⎞<br />
0.0983<br />
= ⎝−0.1106<br />
0.0865 −0.0364⎠<br />
0.0983 −0.0364 0.0847<br />
87