NavList:

Stand by for yet another of Dave’s Dodgy Descriptions for you to blow holes in.  I think this scientist has got mixed up, which is easily done.  I’m going back over 20 years now, but I don’t think satellites in space have inertial systems onboard, but as Antoine says most modern land sea and air inertial navigation systems in the ‘User Segment' will be hybrid systems with GNSS or other monitoring using the dreaded Kalman filtering (now that really does scramble your brain).  Often the first thing to diverge is inertial heading, and as Antoine said, after that the inertial transducer inputs, however accurate in themselves, are applied in the wrong direction.  Pre GNSS, the RAF Vulcans and Victors used FMA (Fix Monitored Azimuth) to tweak the heading of Blue Steel Missile inertial platform.  The Jaguar might have done something similar; Antoine might know.  The Nimrod I believe placed great faith in a good steady take-off run on an accurately known runway heading to give its IN platform a good start.

It’s also possible that a single satellite might start to go ‘wonky’ some time before it’s picked up by one of the ‘Ground Segment’ monitoring stations of which there were only about three worldwide for GPS.  This was one of the reasons why totally GNSS reliant air navigation for non-military traffic took a while to be accepted, and GNSS final approaches and landings were the last to be accepted.  Here again, there might be scope for misunderstanding. I think it’s the ‘User Segment’ that monitors one satellite against another, or to be more accurate, one batch of four satellites against another to isolate the duff satellite, not one satellite in space to another.  This allowed GNSS to be cleared for en-route air navigation at least, and put blokes like me out of a job.  It’s called RAIM (Receiver Autonomous Integrity Monitoring), and you can read all about it on Wikipedia.  For GNSS approach and landings you need much more critical monitoring of which I know nothing. 

Stand by for yet another of Dave’s Dodgy Descriptions for you to blow holes in.  I think this scientist has got mixed up, which is easily done.  I’m going back over 20 years now, but I don’t think satellites in space have inertial systems onboard, but as Antoine says most modern land sea and air inertial navigation systems in the ‘User Segment) will be hybrid systems with GNSS or other monitoring using the dreaded Kalman filtering (now that really does scramble your brain).  Often the first thing to diverge is inertial heading, and as Antoine said, after that the inertial transducer inputs, however accurate in themselves, are of little use.  Pre GNSS, the RAF Vulcans and Victors used FMA (Fix Monitored Azimuth) to tweak the heading of Blue Steel Missile inertial platform.  The Jaguar might have done something similar; Antoine might know.  The Nimrod I believe placed great faith in a good steady take-off run on an accurately known heading to give its IN platform a good start.

It’s also possible that a single satellite might start to go ‘wonky’ some time before it’s picked up by one of the ‘Ground Segment’ monitoring stations of which there were only about three worldwide.  This was one of the reasons why totally GNSS reliant air navigation for non-military traffic took a while to be accepted, and GNSS final approaches and landings were the last to be accepted.  Here again there might be scope for misunderstanding. I think it’s the ‘User Segment’ that monitors one satellite against another, or to be more accurate one, batch of four satellites against another to isolate the duff satellite, not one satellite in space to another.  This allowed GNSS to be cleared for en-route Air navigation at least and put blokes like me out of a job.  It’s called RAIM (Receiver Autonomous Integrity Monitoring), and you can read all about it on Wikipedia.  For GNSS approach and landings you need much more critical monitoring of which I know nothing. 

By the way, the satellites in orbit rarely navigate themselves although they can be tweaked if the get a long way off station.  They follow known orbits in space. It’s all down to that Kepler chap, and their positions are predicted to your receiver well in advance.  Then the ground stations monitor each satellites position and tells it how far it’s off its predicted position and what its clock error is.  The satellite then signals these corrections back to user receivers by way of its ‘signal from space’.  
Usual caveats apply, I might just have dreamt all this up after too much cheese at bed time.   DaveP