From: Phil Karn <karn@unix.ka9q.ampr.org>
Date: Fri, 5 Feb 93 00:02:23 PST
To: jpp@markv.com
Subject: spread spectrum synchronization
Message-ID: <9302040757.AA02215@unix.ka9q.ampr.org>
MIME-Version: 1.0
Content-Type: text/plain


Synchronization is indeed one of the more complex parts of a spread
spectrum system. But there are standard, well studied ways to do it,
and any of several textbooks have good descriptions. It's more tedious
than difficult. A good job for VLSI.

The "gold codes" you mention are simply one family of spreading
sequences.  There are others.  You'd search for signals that use them
in the same way you'd search for any kind of spreading sequence -- by
searching the sequence space until you find a correlation between the
incoming signal and your local copy of the spreading sequence. You
detect this by the sudden increase in narrowband signal energy seen at
the output of the despreader around the nominal carrier frequency when
you're within a chip of the correct time. You then lock onto and track
the spreading sequence with a circuit very similar to a phase locked
loop.

Practical spread spectrum systems aid this process either by
synchronizing the spreading sequence to a common clock source, or by
using a small enough spreading sequence to make searching the entire
sequence feasible.

In our CDMA cellular system, we use both approaches and several
ordinary LSFR PN sequences (not Gold codes). There's a "short PN" code
(2^15 chips long) common to all cells that can be found by a mobile
within a few seconds of being turned on, and an additional per-user
"long code" (2^47-1 chips) that is synchronized with the aid of GPS
receivers at the cells. Both operate at the same chip rate (1.2288
Mhz).

The gory details can be found by anonymous FTP to lorien.qualcomm.com,
in /pub/cdma. Read the "overview" document first.

Phil