On Mar 09, 1996 at 00:24, Paul Edwards of 3:711/934.9 wrote:

 PE> Come on David, you can't drop bombshells without giving us the
 PE> juicy bit.

Well .. in response to a request I made recently, I was sent a PostScript
formatted copy of a document entitled:

    "ITU-T V.34 (09/94)

    "Data Communication Over the Telephone Network

    "A Modem Operating at Data Signalling Rates of Up to 28,800 bit/s
    For Use on the General Switched Telephone Network and on Leased
    Point-to-Point 2-Wire Telephone-Type Circuits"

As I said, I don't really have this document.  :-)

 db>> but from what I have seen, there's no canonical statement about the type
 db>> of trellis coding that should be supported in order to claim V.34
 db>> compliance - 'cept of course that it's gotta be one of the three above.
 PE>
 PE> So that's the canonical statement?  The spec IS the spec!  What
 PE> more do they need to say?

Here are some samples:


    "This modem is intended for use on connection on general switched
    telephone networks (GSTNs) and on point-to-point 2-wire leased
    telephone-type circuits.  The principal characteristics of the modem are
    as follows:

    ...

    "e)     trellis coding for all data signalling rates;"

        - s1, "Scope" (p. 1)


    "trellis encoding: A method for improving noise immunity using a
    convolutional coder to select a sequence of subsets in a partitioned
    signal constellation.  The trellis encoders used in this Recommendation
    are all four-dimensional (4D) and they are used in a feedback structure
    where the inputs to the trellis encoder are derived from the signal
    points."

        - s3, "Definitions" (p. 3)


    "The encoder shall be selected by the receiving modem during Phase 4 of
    the start-up procedures specified in 11.4 or 12.4.  The following
    encoders are available:

    "- 16-state rate 2/3 (Figure 10);

    "- 32-state rate 3/4 (Figure 11);

    "- 64-state rate 4/5 (Figure 12).

    "For the 32-state encoder, the input bit Y3(m) is not used.  For the
    16-state encoder, the input bits Y4(m) and Y3(m) are not used."

        - s9.6.3.2, "Convolutional Encoder" (pp. 24-25)


    "Modulation Parameter (MP) Sequences are exchanged between modems
    during start-up and rate renegotiations and contain modulation parameters
    to be used for data mode transmission.

    "Two types of MP sequences are used in duplex mode.  Type 0 contains ...
    trellis encoder choice ..."

        - s10.1.3.9, "Modulation Parameter (MP) Sequences" (p. 33)

 PE> There is a flag, the flag must have one of three values, what's the
 PE> problem?

Four (4) values - but one's reserved by the ITU;  the other three are of
course the three convolutional encoders mentioned above.  As it says in
table 20/V.34, "Receiver requires remote-end transmitter to use
selected trellis encoder."

 PE> There's probably a statement about what to do if one person has 64 and
 PE> the other has 16.  Probably something about 16 needing to be supported
 PE> as a common minimum, or that the modems can do without it altogther
 PE> or something.

Errmmm .. nope, not that I see.  But as I said, I haven't been through here
in detail.

 PE> More likely, 64 means you support UP TO 64, ie you also support 16 and
 PE> 32.

That would be a reasonable assumption.  I'll have a bit more of a read of
s11.4 and s12.4 and see what they say about trellis encoder negotiation.

[...later...]

Y'know, the C spec is more explicit than this - fewer ambiguities.  They
must assume that one has access to and has read previous TCM specs (like
V.32 and V.32bis).  Not clear at all.  :-(

Cheers..


    - dave
    d.begley{at}ieee.org

---