Stochastic model considering individual satellite signal quality on GPS positioning

Simple stochastic model is normally used in GPS positioning by making assumption that all GPS observables are statistical independent and of the same quality. By the above assumption, similar variance is assigned indiscriminately to all of the measurements. A more detail stochastic model considering specific effects affecting each observable individually may be approached such as the ionospheric effect. These effects relate to phase and amplitude measurements in satellite signals that occur due to diffraction on electron density in the ionosphere. This is particularly relevant to those regions frequent with active ionospheric event such as equatorial and high latitude regions. A modified stochastic model considering individual satellite signal quality has been implemented which based on the computation of weights for each observable. The methodology to account for these effects in the stochastic model are described and results of experiments where GPS data were processed in relative positioning mode is presented and discussed. Two weighting parameters have been used in the experiment: elevation angles and tracking error variance in the GPS receiver. The results have shown improvement of 10.3% using the elevation angles as weighting parameters and 11.5% using the tracking error variance as weighting parameter.