GPS Selective availability turned off

Lev Bishop lev.bishop@queens.oxford.ac.uk
Wed, 3 May 2000 02:20:33 +0100 (BST)


I guess my article on GPS for cavers is now a little out-of-date...
http://users.ox.ac.uk/~quee0367/gps2.html

To answer your (Wookey's) question:

"Long-baseline" is a term with more than one meaning. For carrier-phase
techniques, it generally refers to the cutoff between using different
types of processing on the data (eg "float", "fixed-static",
"quick-static" and so on). For these purposes, cutoffs of around 10-20km
seem to be recommended, depending on various factors (number of satellites
being tracked, L1/L2 receiver versus L1 only, etc). Removal of SA will
make no difference at all to this kind of survey. Well, perhaps it will in
one way - the SA was actually implemented as both "dither" (intentional
timing degradations of the transmitted signal) and "epsilon" (intentional
degradation of the transmitted ephemerides). I don't believe the epsilon
was ever used, however (but I may be wrong here). If it were implemented,
then the ephemeris errors would have showed up as 2nd-order effects in
differential calculations. This would only be relevant to realtime GPS as
for post-processed work it's possible to download precise ephemeris data
from the net anyway. Either way, it's a negligable difference.

The other interpretation of "long baseline" is the length of baseline at
which you're better off using the transmitted ionospheric modelling
parameters rather than assuming you have the same ionospheric conditions
at both the base- and moving-stations. I think this cutoff is at roughly
250km, depending on level of ionopheric activity. Since ionospheric errors
are the largest remaining source of error for single-frequency GPS now
that SA is gone, applying DGPS corrections (in the most straightforward
way, anyway) is unlikely to be useful for longer baselines than this. The
amount of ionospheric delay does vary somewhat however. We are currently
at a sunspot maximum (see recent aurora borealis in northern england?), so
the errors will be larger than usual. Also the (uncorrected) ionospheric
delay varies from 40-60m (95%) in the day to 6-12m (95%) at night. So
perhaps we'll get better GPS performance at night :-) (Those figures are
for uncorrected ionospheric delay - I don't know what happens to the
errors once you've subtracted the modelled delays.)

One change which will result from the removal of SA is that DGPS
corrections will not need to be sent at such a high rate, since the
bandwidth of ionospheric, tropospheric, SV clock, etc errors is much less
than the SA-induced errors. I believe that the DGPS coastal beacons
already have a 50bps standard, in anticipation of deactivation of SA,
whereas before 100 or 200bps was necessary (transmitting rate-of-change
information as well as corrections).

I think only way to get a lot of benefit from DGPSip in the post-SA world
is to turn it into a so-called "Wide Area DGPS" (WADGPS) system. In a
WADGPS system you observe the GPS signals at a variety of fixed sites
worldwide and you come up with a model for the atmosphere (iono- & tropo-)
which is valid worldwide. You transmit the parameters of this model to
your roving station, which can then remove the atmospheric errors at it's
own location. This is the kind of thing OmniSTAR has been doing with their
Virtual Base Station (VBS) service. I'm not sure how many sites would be
necessary for this, but the OMNIstar system uses about 70 (and the IGS
uses 73 in its post-processing) to get worldwide coverage. The removal of
SA would make adding this kind of thing to DGPSip a lot easier because the
errors will change so much more slowly, giving you more time to collect,
process and distribute the data. With SA the corrections needed to be
updated roughly every minute I think that without SA the data will remain
valid for 10minutes or more. Also, the rapid changes in errors with SA
meant that you had to either take all your measurements simultaneously, or
at least be able to extrapolate them all to a common time, but this should
matter less now. Also, with SA measuring the contribution of the
ionosphere meant you needed dual-frequency equipment, but now that SA is
gone you should be able to get by without this added complication. I guess
we'd need a fair few raw-data GPS units scattered around the place, and
some protocols and datapaths for collecting the data in one place, where a
computer would make a solution to distribute back to users. The users
could still use normal consumer gps units as they'd connect to a laptop
with a net connection. The laptop would get a rough fix from the gps, use
that position in a model of atmospheric parameters, and synthesise an
RTCM-104 correction message to pass back to the GPS.

For post-processed work, you can already download ionospheric maps based
on IGS data. The IGS isn't yet distributing an official Ionospheric data
product to go with the precise ephemerides they make available, but some
of its members are doing so. For example CODE is doing this. There's a
standard format for such data, called IONEX. CODE also tells you each
satellite's differential code bias (DCB) which is I think needed for
relating dual-frequency (as measured at the reference stations) to
single-frequency results. With a raw-output GPS receiver you should be
able to log data and get fixes limited only by receiver noise and
multipath.

In fact, since you can download 2-day predictions of ephemerides, and
ionospheric parameters from these places, you ought to be able to load the
parameters into a laptop in the morning and get real-time receiver-limited
fixes all day in the field without receiving any corrections! Now we just
need free software to do the processing...

Lev

On Tue, 2 May 2000, Wookey wrote:

> Indeed. We (the BCRA CSG) were working on setting up a european DGPSip
> station to supply differntial GPS info over the internet. This worked
> usefully over a 3000mile baseline, and would be a useful adjunct to the
> two US stations. All you need is a permanent net connection within 150
> miles of a DGPS transmitter and about 400 quid-s worth of DGPS
> receiver+transmitter.
> 
> This would have provided better GPS services for cavers anywhere in
> europe (so long as they could get access to the net), however this change
> may have made the idea largely redundant as non-SA GPS is probably good
> enough for most cave-location purposes. I'm waiting to find out what the
> expected differential numbers are under the non-SA regime, especailly
> under long-baseline condiditions. Anyone here know?

 ----------------------------------------------------------------
| Lev Bishop            The Queen's College, Oxford, OX1 4AW     |
| Physics student                                                |
| mail: lev.bishop@queens.oxford.ac.uk (quee0367@sable.ox.ac.uk) |
| http://users.ox.ac.uk/~quee0367/      ICQ# 21136345            |
 ----------------------------------------------------------------