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 ambiguity resolution 3<br />
<strong>The</strong> problem of <strong>integer</strong> ambiguity resolution has drawn a lot of attention in the past<br />
decades. Many ambiguity resolution algorithms have been published. Not all of these<br />
algorithms will be discussed here; only a brief overview will be given. <strong>The</strong> first step<br />
of ambiguity resolution is <strong>integer</strong> <strong>estimation</strong>. <strong>The</strong>refore, section 3.1 starts with the<br />
definition of admissible <strong>integer</strong> estimators. For the quality description <strong>and</strong> <strong>validation</strong> of<br />
the estimators, their distribution functions are required. <strong>The</strong> distributional properties of<br />
the float <strong>and</strong> fixed ambiguity will be given in section 3.2, where it is also shown how the<br />
probability of correct <strong>integer</strong> <strong>estimation</strong>, the success rate, can be approximated. For the<br />
purpose of <strong>validation</strong>, the parameter distribution of the ambiguity residuals is required.<br />
This distribution function <strong>and</strong> its properties will be given in section 3.3. In section 3.4<br />
it is shown how the quality of the fixed baseline estimator can be expressed. Section 3.5<br />
gives an overview of currently available methods for the <strong>validation</strong> of the fixed ambiguity<br />
solution <strong>and</strong> their shortcomings. Finally, in section 3.6 a completely different approach<br />
of ambiguity resolution is described, namely the Bayesian approach.<br />
3.1 Integer <strong>estimation</strong><br />
Any <strong>GNSS</strong> observation model can be parameterized in <strong>integer</strong>s <strong>and</strong> non-<strong>integer</strong>s. This<br />
gives the following system of linear(ized) observation equations:<br />
y = A a + B b + e<br />
m×nn×1 m×pp×1 m×1<br />
m×1<br />
(3.1)<br />
where y is the GPS observation vector of order m, a <strong>and</strong> b are the unknown parameter<br />
vectors of dimension n <strong>and</strong> p respectively, <strong>and</strong> e is the noise vector. <strong>The</strong> data vector<br />
y usually consists of the observed-minus-computed DD phase <strong>and</strong>/or code observations<br />
on one, two or three frequencies <strong>and</strong> accumulated over all observation epochs. <strong>The</strong><br />
entries of the parameter vector a will then consist of the unknown <strong>integer</strong> carrier phase<br />
<strong>ambiguities</strong>, which are expressed in units of cycles rather than in units of range. It is<br />
known that the entries are <strong>integer</strong>s, so that a ∈ Z n . <strong>The</strong> remaining unknown parameters<br />
form the entries of the vector b. <strong>The</strong>se parameters may be the unknown baseline<br />
increments <strong>and</strong> for instance atmospheric (ionospheric, tropospheric) delays, which are<br />
all real-valued, i.e. b ∈ R p . <strong>The</strong>se real-valued parameters are referred to as the baseline<br />
parameters, although the vector b may thus contain other parameters than only the<br />
baseline components.<br />
27