Avionics were designed with the opposite assumption. When a high accuracy location source provides a location (GNSS or DME/DME) the position is updated and it's assumed that could mean a significant jump for the INS.
Practically that was what happened all the time before GPS. You would fly for a few hours over the ocean with no ground based reference, having a reasonable but not perfect INS location. Then get close to the coast where a DME/DME fix was done which updated the INS position as a big jump of up to a few miles.
Filtering GPS updates that are too far "off" the INS state would be an almost opposite design to the original assumptions that DME and GNSS are highly accurate.
Re:Flightradar24's new GPS jamming map
by randomcarbloke on Friday March 22, 2024 @12:27PM
(#39792256)
some amount of IRS drift should be ~predictable, an anticipated deviation from GPS could be used as a measure of trust for GPS
I understand that current civilian aircraft wasn't designed with that in mind though.