Path: utzoo!attcan!uunet!snorkelwacker!usc!ucsd!pacbell.com!pacbell!hoptoad!hsfmsh!daemon
From: tnixon@hsfmsh.UUCP (Toby Nixon)
Newsgroups: comp.dcom.modems
Subject: Re: Want info on Trellis encoding for V.32
Message-ID: <3685@hsfmsh.UUCP>
Date: 30 Jul 90 01:54:28 GMT
Sender: daemon@hsfmsh.UUCP
Organization: Hayes Microcomputer Products, Inc.  Norcross, Georgia
Lines: 58

In article <1397@wet.UUCP>, Eric P. Scott writes:

- What does it do?  Do all V.32 modems support it?  If not, how
- serious is not having it?

Trellis coding is a technique whereby additional bits are 
transmitted with the user data, calculated based on the preceding 
data, so the that receiving modem can "predict" what at least one of 
the transmitted bits will be and therefore have a better chance of 
properly deciding what the received "symbol" actually represents 
(this is greatly simplified).  When trellis coding is enabled in 
V.32 modems, one can conceive of them as operating at 12,000 bits 
per second, with one of the bits in each "symbol" (2400 symbols are 
transmitted per second) being a "coding" bit (so the actual user 
data throughput is 9600 bit/s).  The number of points in the 
constellation is expanded from 16 to 32 as a result, but at the same 
time the ability to "anticipate" what the coding bit "ought" to be 
allows the modem to make a better guess at the subsequent symbols.

When V.32 modems were first shipped (it has been a standard size 
1984), some did not have trellis coding (in fact, there are some 
that only support 4800 bit/s).  At the present time, however, 
I would guess that all V.32 modems support trellis coding.  It is so 
ingrained, as a matter of fact, that the new proposed V.32bis 
standard (which adds 7200, 12000, and 14400 bit/s to the V.32 speeds 
of 4800 and 9600) mandates trellis coding at all speeds except 4800.

The main problem with operating without trellis coding is that 
you'll experience more line noise.  Trellis coding allows the modem 
to tolerate about 3.5 dB additional signal-to-noise ratio than a 
modem without trellis coding, at the same bit error rate.  That can 
be quite significant.  On some alternative long-distance networks 
that use tandem ADPCM links (adaptive differential pulse code
modulation is a technique for squeezing twice as many calls onto a 
digital trunk) you'll find that a non-trellis-coded V.32 modem 
probably won't work AT ALL at 9600, but would be limited to 4800.  
Operation on ADPCM links (which are common on intercontinental 
circuits) is one of the reasons the Group 3 Fax standards committees 
have been anxious to adopt a standard (the proposed V.17 standard) 
which will add trellis coding to Group 3 fax at 7200 and 9600 (the 
V.29 standard currently used at those speeds doesn't have trellis 
coding and won't work well on tandem ADPCM links).

Anyway, that's a lot of words to basically advise you that if I were 
you, I wouldn't even consider buying a V.32 modem without trellis 
coding!  It would interwork with all other V.32 modems, but you'd 
pay a real performance penalty.

	-- Toby

-----------------------------------------------------------------------------
Toby Nixon, Principal Engineer     Fax:    +1-404-441-1213  Telex: 6502670805
Hayes Microcomputer Products Inc.  Voice:  +1-404-449-8791  CIS:    70271,404
Norcross, Georgia, USA             BBS:    +1-404-446-6336  MCI:       TNIXON
                                   Telemail: T.NIXON/HAYES  AT&T:     !tnixon
UUCP:   ...!uunet!hayes!tnixon     Internet:        hayes!tnixon@uunet.uu.net
MHS:    C=US / AD=ATTMAIL / PN=TOBY_L_NIXON / DD=TNIXON
-----------------------------------------------------------------------------