The Precise Point Positioning (PPP) has become a powerful satellite positioning technique which nearly equals performances provided by advanced relative positioning techniques. Exploiting the growing availability and quality of IGS products (satellite orbit and clock products), the PPP technique can now provide a centimetre level solution in static mode and a decimetre level in kinematic mode. However, the PPP technique still presents some weaknesses. In order to reach a high precision level, it requires a significant convergence period which can typically reach 30 minutes under normal conditions. Moreover, the PPP seems to be especially sensitive to ionospheric scintillations effects which involve signal amplitude and phase variations of GNSS signals. These weaknesses still limit the use of the PPP technique in the frame of some specific and demanding applications (agricultural industry, airborne mapping, etc.). The goal of our research project is to develop new data processing strategies attempting both to make the PPP technique more reliable under ionospheric scintillations and to optimize the PPP convergence time. The project is composed of several workpackages aiming to improve the mentioned current PPP weaknesses with specific strategies. One of the workpackages is devoted to the impact of satellite geometry on PPP performances. Ionospheric scintillations are susceptible to reduce the number of tracked satellites which degrades the quality of satellite geometry. Based on an analytical development, we first attempt to figure out what types of satellite geometry can be harmful. Then, we discuss about the improvement of the satellite geometry quality involved by the combined use of GPS and Galileo and its benefits in the frame of the PPP. Another workpackage is related to the weighting scheme. Based on an iterative least-square adjustment, the PPP algorithm requires the definition of a stochastic model composed of an observation covariance matrix. Usually, this matrix is chosen as diagonal with zero covariances assuming that correlations between observations can be neglected. In particular, our project aims to study the validity of this stochastic model for the PPP in order to determine whether tuning the weighting scheme of the stochastic model can improve the PPP performances. By exploiting spatial analysis techniques, we try to characterize the spatial auto-correlation between GNSS observations, considering the signal-to-noise ratio as the main observable. From the results of these experiments, we will discuss about the spatial correlation between GNSS observations both under normal conditions and ionospheric scintillations.