NavList:

   

Reply

Paul Hirose wrote:
> [...]
> I'm skeptical about this SA eliminator program.  The developer says it
> has .0001 minute accuracy?  Maybe it does, but I'll bet this is an
> extreme figure, based on occupying the site for an impractical length
> of time.  I'd like to see his test results.
> [...]

Yes, the problem is: you need not only a lot of fixes for averaging,
they also must be adequately spaced in time.

Improving a result by averaging individual measurements assumes that
the errors in these measurements are randomly distributed so that
they have a tendency to cancel out in the average.
However, the errors induced by SA are varying only _slowly_, so that
two fixes obtained within a few seconds have more or less the same
error and nothing is gained by averaging them. The well-known rule
about the error of the average decreasing as the square root of the
number of the measurements averaged does not apply for these
'correlated' measurements.

So you have to wait a while until SA has produced a 'new' error to
obtain a new, uncorrelated fix.

A rough estimate:
Let's say the average error of a fix is 50 m (as a _very_ rough number).
That's about 0.03 min of arc. Getting down to 0.0001 means an
improvement by a factor 300. _If_ your fixes are uncorrelated, that
means you need 300^2 = 90000 fixes for averaging. In order to have your
fixes uncorrelated, you have to wait say four minutes between fixes.
That results in a total measurement duration of 90000*4 minutes or
250 days.

And that's assuming there are no systematic errors (like deficiencies
in the atmospheric delay models etc. etc.)

Any comments on this estimate?


Anyone who would like a closer look at some averaging results
may want to visit
           http://www.erols.com/dlwilson/gps.html


Kind regards,
  Th. Schmidt
--
schmidt@hoki.ibp.fhg.de                 > This is a standard disclaimer
<
                                        >      This is a fancy .sig
<
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--=-=
=-=  TO UNSUBSCRIBE, send this message to majordomo@ronin.com:     =-=
=-=       navigation                                         =-=
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--=-=

   

Reply