[time-nuts] I thought GPS repeated every 12 hours (-2 minutes)

Michael Wouters michaeljwouters at gmail.com
Thu May 26 04:01:50 EDT 2016


> Multipath on GPS normally requires a couple of things:
>
....
> 3) The signal path length has to be close enough that the normal firmware does not reject
> the solution.
>

Expanding on this a bit, because it's relevant to the "aircraft
causing multipath" question, the pseudo random noise ranging method
allows discrimination in the time domain against sources of multipath
further away than 300 m at worst, and much better in practice.

At the L1C 1 MHz chip rate, the correlation function for the received
PRN is zero more than 1 us == 300 m away. In practice, you're
digitizing much faster than 1 MHz and you can use a more closely
spaced set of correlators, perhaps improving by a factor of 10.

This doesn't work though when a reflection is the only signal you are
seeing,  as in Bob's point (1).

(The effect of multipath when there is a line of sight signal also
visible is to distort the correlation function, which is triangular in
the ideal case. This introduces errors into the interpolation used to
improve the resolution of the PRN ranging.)

Cheers
Michael


More information about the time-nuts mailing list