Satellites selection can be performed based on constellation geometry, ranging-signal quality, or both. For applications where multipath and blockage are not major problems, Signal-in-Space (SIS) user range error varies less significantly, which is the case for many applications most of the time. Constellation geometry is then the key criterion for satellite selection. This paper focusses on the constellation-geometry aspect of satellite selection. A review of existing theoretical analyses for Dilution of Precision (DOP) minimization and existing satellite selection methods is initially given. An extended analysis of existing methods and discussion of their pros and cons is then provided followed by four simulated GNSS constellations to evaluate the methods. Finally, a summary of the existing methods and one modified method regarding performance in DOP values and the corresponding computational efforts is presented.