@layman,
The issue here is decades old, and was resolved long ago.
Hafele and Keating made the first attempt to measure clock retardation due to motion.
There were 3 clocks involved. All three started out sitting on a table at a Naval station in Maryland.
Then two of the clocks were put on planes.
Then each plane was accelerated, IN OPPOSITE DIRECTIONS to identical cruising speeds (say 500 mph)
Eventually each plane circumvented the globe and returned to the naval station. All three clocks were reunited on the table top. And, as it turned out, each clock had recorded a different amount of time elapsed during the interim. No two had the same reading in this respect.
This was contrary to the original expectations and could not be reconciled by trying to apply SR.
The only frame which could be used to correctly "predict" (actually postdict in this case) the readings actually recorded was the ECI. It had to be used as the preferred frame from which to interpret the data actually gathered empirically.
It's the same with the GPS