[time-nuts] PN sequence generation using GPS

[Hal Murray]

I'm not familiar with FHSS or DSSS, but pseudo random sequences from LFSRs 
are used as scramblers on serial links.  The idea is to avoid long strings of 
0s that will provoke clock recovery troubles.  The transmitter XORs a pseudo 
random sequence with the data stream and the receiver does the same to get 
the data back.

The problem is to keep both ends in sync.

There is a good section on scramblers in Lee and Messerchmitt: Digital 
Communication, page 440.
I don't know how good that book is.  It's one I picked up years ago.  I've 
been happy with it.  Mine says copyright 1988, 4th printing 1992.

There are two types of scramblers: Frame Synchronous and Self Synchronizing.

With a Frame Synchronous setup, you get a reset pulse from someplace else.  
(magic)  It's usually  a start-of-packet signal or a specific bit pattern 
used as a frame marker, anything both ends can agree on.  The idea is to get 
back in sync occasionally in case you happened to get out of sync.

So the questions are:
  Can both ends get the same reset pulse?
  Are the clocks at both ends running at the same speed?

I'd expect it to work as long as both GPS systems are happy.

You will probably have to correct for propagation delays, that is delay the 
receive clocking by X ns.


The self synchronizing scramblers probably aren't appropriate for SS work.  
(But they might work fine if you know how to use them.  I don't see it.)

The idea is to feed the data bit on the wire into the LFSR rather than the 
output of the LFSR.  If the data is received correctly, a garbage scrambler 
on the receive side will get in sync after N bits.  If you get a single bit 
error on the wire, it will make a multi-bit error in the unscrambled data 
stream.  That error pattern will be the polynomial used for the LFSR.  (It's 
easy after you see it.)




-- 
These are my opinions, not necessarily my employer's.  I hate spam.