Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: [EFM] EFM Requirements

I just reviewed the last two months of this exploder and I would like to
focus on some of my favorite questions for EFM Copper as I think the Optics
discussion is more focused..

Why even bother talking about DOCSIS?

There is no technical relationship to an IEEE 802.3 device.  A DOCSYS modem
is effectively a radio based QAM modem running on coax specifically designed
to be backward compatible with the existing North American cable TV systems.
DOCSYS is extremely limited in expandability by definition in order to
maintain compatibility and is not an '802.3' device.  DOCSYS uses an IEEE
802 specified MAC.  Otherwise it has no relationship to Ethernet.  What am I
missing on this?

Let us please try to focus on a topology we are allowed to implement in our
humble IEEE standards forum.

We a huge list of work required just for copper.

1) We need to choose a modulation method for copper wire.  (ADSL, VDSL V.dmt
or QAM or a combination)

2) The modulation must be spectrally compatible with other technologies.
(T1E1) or the owners of the wire will never let us use them if we trash
other applications on the telephone cable.  This limits us to pre-specified
symmetry of the data rate on the subscriber line.  Some suggest we eliminate
CO and CPE devices and use Answer and Originate to determine the symmetry of
the data between the user and the network.  But even this would blow the
T1E1 spectral compatibility compliance, so we can't do it.  It seems we are
stuck with a defined asymmetric data rate unless we want to all adopt the
Swedish spectral plan instead of 998 at previously agreed in ITU-T and T1E1.
To do this, we would have to get the ITU-T and T1E1 to toss out the last
three years of committee work and revisit spectral planning.  This would
only happen if an Act of Congress were to be passed outlawing a predefined
spectral compatibility template.  In regulatory the opposite seems to be

3) For reach we could use ADSL, that is 2Mbit (downstream) at 17K feet (26
AWG w -140 dBm noise) or 20 K feet on 24 AWG, for data-rate we could use
VDSL using QAM or V.dmt versions of VDSL to obtain 25-30 Mbit at 3K feet and
10-15 Mbit at a maximum distance of 6K feet.

4) We could specify a combination ADSL, VDSL data-pump to yield maximum
reach with a close-in high data-rate.  This would be an easy combination to
build if the selected VDSL pump strategy was dmt and very complex if the
selected strategy was QAM.  Is there any market advantage for an ADSL / VDSL
Phy Ethernet combination data-pump?

5) We must decide if and how we are going to select between QAM of dmt.  It
is commonly believed QAM is a less expensive modulation strategy and V.dmt
used on ADSL is more robust and expensive to build.  The billion-dollar
question is:  Does the small ever-decreasing marginal cost difference
between QAM and V.dmt justify the small marginal increased reach V.dmt has
over QAM?  How do we plan to solve this question?

6) As a practical issue should we consider splitting the EFM group into
three sub-groups, concentrating on a Phy for copper, a Phy for optics and a
MAC for the EFM topology?

7) Actual compatibility with HPNA is not required.  HPNA band-pass filters
can be placed on the line.  You can't use HPNA and VDSL on the same wire
without filters to protect each other because they use the same transmission
bands.  You just can't have everything.  A splitter works great for HPNA.
The fact is if you have HPNA you might require a truck-roll for VDSL.

Thanks, Daun.