🚀 go-pugleaf

In <2041@deadmin.ucsd.edu> gumby@ucsd.edu (Henry Ptasinski) writes:

>The CROSS-CORRELATION between two PN sequences can have  
>large peaks at some values of tau, even if the PN codes are not identical.   
>Careful selection of the PN sequences is required in order to have a set with
>good cross-correlation properties.

Of course you're right.  The GPS PN sequences have been carefully selected
so that there are 37 unique codes that are supposedly especially easy
to acquire.

>gsulliva@enuxha.eas.asu.edu (Allen Sullivan) writes:

>I would expect there to be problems in discriminating among codes if very
>short sublengths were tested for. For example, 1100111000 probably occurs
>in all the codes. So once "trial correlation" happens, longer/entire pattern
>matches must be done. What is a useful sublength to use, to adequately avoid
>false trial correlations?

Since the code repeats 1000 times a second, why wouldn't you use the whole
thing?  A particular receiver we're using here takes 5 minutes to "search
the sky" for each satellite in order (starting with PRN 6) when it has no
almanac information. It takes 18-20 minutes for acquisition if PRN 6 is in
view. (Since PRN 6 is no longer in service, acquisition times of over an
hour are not uncommon.) This compares to 2-3 minutes if the receiver has a
current almanac and knows which satellites are visible. While I don't have
access to the internals of the receiver, I suspect it uses the entire 1023
bits of the C/A code.
-- 
Andy Arkusinski					arkusinski@benzie.si.com
Smiths Industries				arkusinski@esseye.si.com
Grand Rapids, MI