There are no definitive requirements to say the satellite clock
frequency must be with certain percentage of the ground clock
frequency. This is explained by the manufacturers of GPS receivers.
http://www.gpsinformation.org/dale/theory.htm
For example, the satellite clock can all tick at 1.000GHz +/- 0.5Hz
while the ground clock can tick at 577.9874MHz +/- 10.67MHz.
When your receiver receives almanac information from four satellites
with each giving its time, altitude, longitude, and latitude, you can
form a set of four equations with four unknowns. The unknowns are
your satellite time (nothing to do with ground time), your altitude,
your longitude, and your latitude.
Professor Andersen knew this, but trying to promote the nonsense of SR
and GR, he proposed a fairy-like mysterious function to the GPS where
this mystic function requires the synchronization of the satellite and
ground clocks. Professor Roberts recently understood this. In doing
so, you also proposed a top-secret military function built into the
GPS that also requires the synchronization of the satellite and the
ground clocks. It becomes a faith issue. However, I call the bluffs
of these professors. I do not buy the mysticism they try to promote
for SR and GR. <shrug>