GPS and security

I speculated that the
   government may in fact have designed the satellite systems so that
   they could be told to do several things in case of national emergency:

   *	Shut down (probably possible; they may have actually mentioned
   this on the show).  Problem is that lots of friendlies may grow to
   rely on the data for life-critical things, like guiding commercial
   airliners.

It's worse; the military relies on GPS. During the Gulf War large numbers of
combat vehicles could rendezvous with high accuracy and no radio contact in
the dark of night; staggeringly important in limiting fratricide as well
(though not as well as one might have liked...)

   *	Shut down normal transmission and begin strongly encrypted
   transmission.  No mention of this; apparently, the satellites were
   originally designed with some sort of weak system that made the data
   difficult to use for high-accuracy purposes, but that's been defeated
   (by the FAA or someone contracted thereto).

   *	Enter into a bogus-cleartext with encrypted subchannel mode,
   where the plaintext is slyly made to be wrong, but using some
   subchannel encrypted "good stuff" is still available.

Well, this sounds like the "selective availability" (SA) capability, which
still exists and has in fact been turned on. It is defeatable, but only
under certain circumstances (see below).

GPS is a clock-based system; by talking to a constellation of satellites (I
think the minimum is 4, with 5 being desired and 6 being best) and munging
the timestamps received from them, GPS systems can compute their location to
a degree of precision related to the precision of the clocks. SA essentially
truncates a few bits off the low-end of the time stamps and makes them
available in an encrypted form. Without the key to decrypt the low-order
bits, you get location information to within a couple hundred meters or so;
with the low-order bits, within about six feet.

During the Gulf War, the US military was unable to get enough military GPS
receivers (i.e. ones that could decrypt the selectively-available data).
Instead, they bought commercial over-the-counter GPS receivers and turned SA
off. As soon as they got enough SA-enabled receivers, they turned it on.

SA is defeated by something called Differential GPS.  Basically, a ground
station at a fixed location constantly computes its position via GPS,
computes the difference between the GPS location and its known correct
location, and broadcasts the correction factor. A differential-capable GPS
receiver computes its location via GPS and then applies the correction fac-
tor from the nearest differential station. These corrections are obviously
of decreasing accuracy the farther you get from the fixed-position station,
but you can correct for that to some degree once you have an estimate of the
direction and distance from your location to that of the differential
broadcaster; apply corrections iteratively until they converge.

The requirement, of course, is that you be someplace near a differential GPS
station. These stations are maintained by the US Coast Guard and obviously
exist only where there is navigable water (and certainly not everywhere,
since they take time and money to build).

I'm curious if anyone has attempted to crack the encryption used in
selective availability; although Differential GPS solves my most important
issues (coastal navigation) it's be nice to be able to use a handheld GPS
receiver in the middle of Rock Mountain National Park or some such and still
locate yourself on the bloody maps with some precision.

Jason

>The requirement, of course, is that you be someplace near a differential GPS
>station. These stations are maintained by the US Coast Guard and obviously
>exist only where there is navigable water (and certainly not everywhere,
>since they take time and money to build).

Well, yes, the USCG *is* building its first DGPS beacons along the
coasts, for obvious reasons. But there is precedent for the USCG
becoming involved in radionavigation coverage elsewhere in the
US. Several years ago, in conjunction with the FAA, the USCG built
several new Loran-C stations to fill in what had been known as the
"mid continent gap". Now there is good Loran-C coverage over the
entire continental US. If GPS becomes as popular for aviation as
Loran-C has become (at least for helicopter and general aviation, if
not commercial aviation) then I think it quite likely that the same
thing will be done with differential GPS stations.

The only catch is likely to be that if the primary "legitimate" user
of DGPS is aviation, then the DGPS beacons will be sited so as to give
good coverage to aircraft in flight and near airports, not necessarily
good coverage to any user anywhere on the ground. A similar situation
already exists with respect to VORs; get high enough and the entire
country is completely covered, but it's hard to hear one on the ground
unless you're very close to one.

The actual coverage of any DGPS beacon will depend on its frequency
and transmitter power. VHF or UHF would be the most natural places to
put these things for aviation use, and as everyone knows these tend to
be limited to near line of sight, just like VOR. The USCG is putting
its DGPS data on its existing LF (300 Khz range) marine
radiobeacons. These have wider coverage than VHF/UHF, but suffer from
noise and interference problems especially near major cities. Again,
these beacons are optimized for their intended users -- ships at sea,
not users on land.

I should note that the USCG isn't putting up DGPS beacons just to
defeat SA. Even without SA, plain C/A GPS isn't accurate enough for
precise harbor navigation. It just so happens that DGPS takes care of
SA as well as the natural errors they were originally worried about.

An alternative to the USCG-provided DGPS beacons is already provided
via FM broadcast subcarriers as a commercial service. The subscription
price is pretty high, though. They're probably trying to quickly
recoup their investment before the USCG beacons take much of their
market.

Phil