From: pgut001@cs.auckland.ac.nz (Peter Gutmann)
Date: Sun, 8 Nov 1998 11:01:13 +0800
To: cypherpunks@cyberpass.net
Subject: Blind signal demodulation
Message-ID: <91049302221713@cs26.cs.auckland.ac.nz>
MIME-Version: 1.0
Content-Type: text/plain



The October 1993 Proceedings of the IEEE contain a number of rather 
interesting articles on blind signal identification and demodulation, which 
may be described roughly as "demodulation without the cooperation of the 
transmitter (or intended receiver where this might be necessary)".  The first 
article in particular goes into some detail on how to acquire QAM signals used 
in modems, including a neat diagram on p.1919 of a 64-QAM constellation 
through the various stages of acquisition by a blind demodulator.  The blind 
demodulation comletely bypasses the need for an initial training stage, 
acquiring the necessary signal-processing details on the fly.
 
The article finishes with overviews of typical hardware used for blind 
demodulation of QAM signals, including a multi-protocol DSP card with with 8 
320C50's capable of blind demodulation of anything from 24 2400bps signals up 
through 8 V.34 ones, as well as an ASIC for blind demodulation of digital 
cable TV signals.  They also comment that blind decoders for typical voiceband 
signals can be implemented on Pentium MMX/UltraSparc-grade hardware.  This is 
interesting reading, and should lay to rest the UL that high-speed modems have 
some sort of magic immunity to interception which the lower-speed ones don't.
 
Oh yes, the introduction makes the observation that this sort of stuff is 
"rarely mentioned in the open literature".  It's not hard to see why.
 
Peter.