@bwalzer: Er, yes.
If LightSquared's transmitters were so terrible that "Things like intermodulation and IF images" were causing them to radiate outside their allocated band then they wouldn't have a leg to stand on. But by all accounts that's not the case here; no one is saying that LightSquared's transmissions will intrude in an unlicensed or unreasonable way into the allocated GPS band.
The interference of GPS receivers is caused by not defining their receiving band with a filter good enough to reject out-of-band signals that will, if LightSquared start operating, be commonly encountered. All receivers are vulnerable to high power interfering signals causing non-linear responses within the receive chain components. That's why when you design a radio you take a look at the expected operating environment (maybe glance at the frequency allocation tables) and decide how much filtering is going to be needed against legitimate and commonplace adjacent signals. And you are right - guard band practicality is an important factor for the allocating authorities.
It's worth questioning whether today's filtering requirement for GPS is so very different to that theoretically required years ago when the 1525-1559MHz band was allocated to satcoms. To set the historical context for the current debate I think that it is important to analyse the practicality of the previous use of the 1525-1559MHz band for satcoms long before LightSquared came along. And the issues are not generally evident from the online material covering this topic. For example, the last slide at:
shows the frequency allocations and a stylised filter response. It suggests that "Low Power On Earth Satcom Emissions" don't cause interference with GPS, and that "15kW Base Station Emissions" do. However such an analysis is valueless without considering the distance between the interference source and the GPS receiver.
So consider this: a single LightSquared 15kW base station may indeed cause operating problems for a large number of poorly filtered GPS receivers over a wide area. However the low power 2W-ish transmission of a satcoms mobile phone would still have been able to cause problems, just over a much shorter distance (perhaps tens of meters?).
And consider where the satcoms mobile phone might have been used; on a train, on a plane, in a car with SatNav, etc. etc. If they had become as ubiquitous as terrestrial mobiles are today they'd be everywhere all the time. It is perfectly possible that the satcoms phone itself could have caused equally significant problems for GPS just by being physically close to critically important GPS receivers. Anyone who has ever placed a GSM mobile phone near a set of loudspeakers and heard the ticketer-ter ticketer-ter sound it makes will understand my point, especially considering that a set of loudspeakers isn’t designed to be a highly sensitive radio receiver like GPS.
What I seek to show above is that the question of how much guard band to have between GPS and satcoms was just a relevant then as it is now between GPS and LightSquared. It is highly likely that practicality of the chosen guard band was considered long ago by the FCC when the satcom band was first allocated. That consideration ought (hindsight?) to have taken into account a close encounter between a GPS and a satcom phone. That scenario is, from the point of view of a single GPS receiver, not so very different to a more distant encounter with a LightSquared base station. So if the guard band then was considered to be practical, why not now? Maybe the FCC didn't consider such a scenario back then, or maybe they never imagined that everyone and everything would be using GPS receivers for some purpose or other.
By endeavouring to develop an adequate filter LightSquared are seeking to show that the FCC got it right and that the GPS industry are too damned lazy and cheap to do their own jobs properly. The irony of the GPS industry having to license a filter design from LightSquared would be memorable...
However, if the GPS industry is proved right and that filter with adequate out of band rejection can't be build then that really would mean that the FCC got it wrong, and arguably always had got it wrong long, long ago when the 1525-1559 band was first allocated for satcoms.
In both outcomes it's not LightSquared's fault (though you could argue that they should have known better than that). I don't think that it would really be the FCC's fault either. The bands were allocated to satcoms long before anyone thought that every mobile phone, car, etc. would have a GPS receiver in it, so the need for miniaturisation of filters with sharp cut-offs didn't exist. It's not really the GPS industry's fault either. Hardly anyone was actually using the satcoms band. If they had been then we'd have sorted this out years ago. But the solution (whatever it turns out to be) is going to cost a lot of money, and that's always going to come from the customers one way or other.
Some people have pointed out that this is a purely North American problem. But we'd all like our phones (including its GPS) to work properly when we go there.