lml,
"What I was looking for, at a minimum is to first breakdown
& identify, if possible, some distinct components of these architectures
and their "guesstimated" costs.."
Mileage will vary with the architecture that is selected. I know that that
sounds evasive, and it probably is. Truth be told, I don't know what the
actual costs would be for an idealized fiber-to-the-home network (FTTH)
solution today. And an idealized solution, where voice/video/i-net
access/presence were all satisfied at once would probably not come into
being all at once, in the first iteration, anyway.
I would like to be more specific in answering your questions, so let me
see if I can find the specifications and cost justifications that BLS used to
get their live FTTH off the ground. If I can find it I'll post it. It was
published in the hard copy of OSP magazine (one of them), but never
made it to the Web for some reason. Wonder why? [g]
But even here, BLS's solution (btw, I'm not referring to the IFITL roll out
in Atlanta, I'm referring to their "real" FTTH somewhere else in their
territory) is a take-off on analog, not digital. Thus far, anyway.
For now, let's examine two different approaches to incrementalism, one
that has been used by the ILECs, and one which I'll present for
discussion.
---------------
There is the fiber-creep approach being used today by the ILECs, moving
fiber gradually closer to the user, first close enough to support low grade
ADSL, then higher grades of xDSL, and then eventually VDSL hanging
off of a PON scheme.
Another form of incrementalism could exist, the one I'm proposing here:
Whereby, instead of moving fiber closer in successive steps -- an
extremely labor and materials intense proposition -- the fiber is instead
installed directly to the home, first.
Then you implement Internet access and gradually, over time, increase the
number of applications riding over it to include the full magoika of whiz
apps. But you start off, first, with 10 GbE Ethernet supporting Internet
access, primarily, and a limited number of isochronous-like applications.
Those would be voice over IP and two-way video (conferencing) over
IP, and of course, Internet access like I've already stated.
This first pass might support tiered service offerings of Internet Access
(although I'd rather avoid that if possible, but economics and marketing
being what they are) where, at the top, you productize for power users to
have suitable access; telecommuting and video conferencing; some
measured home-based Web-site hosting; all supported by service
guarantees.
At the bottom of the heap you facilitate casual surfing and an
architecture to support municipal applications, say, at a mere 10 Mb/s,
maximum. <s>
-----------
Program and broadcast TV? Sure, eventually. But everyone already has
that, so there is no real need to satisfy that up front. Perhaps you begin with
the enhanced video services first, instead, such as SDV/VoD, and selected
interactive services. But as for basic NTSC TV, users don't need another
variation on the same theme right now, i.e.., the analog, broadcast-only
mode.
And by avoiding analog TV altogether, up front, you are able to keep
overall central site (huge savings there, if you don't have to purchase 1996
technology only to have to rip it out in a couple of years) and premises
costs to a minimum, while providing a service platform that everyone
wants now.
Better yet, you keep the legacy stuff off of the network, entirely. If you
are going to build a greenfield system, try to keep it that way, and refrain
from bringing in an ancient technology that is going to take just as long to
get rid of as it has taken to mature, maybe longer. This also avoids a
major budget line, that which is associated with set top boxes in the initial
stages.
What would the protocol of choice be that supports the undercarriage
here? 10GbE, for starters. Even when shared, 10GbE is approximately
250 times faster than today's shared HFC cable modem when opened up
to a full throttle of 40 Mb/s.
And 10 GbE is approximately 2,500 times faster than HFC
operating at 4 Mb/s, which is more typical of real implementations. Which
is still being pretty generous, on the whole.
And Ethernet would be distributed through modest-sized clusters of
homes, aided by passive splitters along the way to facilitate optimal
clustering.
Upon the next increase in speed to 100 Gb/s Ethernet (and we will get
there within this decade, probably sooner rather than later), the two
comparisons become 2,500 times faster than HFC and 25,000 time faster
than HFC, respectively. And that's if you are using a single wavelength in
a shared mode. That's downstream AND upstream, symmetrically.
By the time anything could be put together to support the platform which
I've just hinted at above, which would take at least a year to begin rolling
out to the first set of homes, 100 Gb will already be in draft form.
There's already a study group working on a framework for 100 GbE now,
in fact, as we type.
As WDM follows the cost curve downward over time, you can begin to
introduce lambda distributions less sparingly, while at the same time
beginning to introduce MPEG video in selected locations.
Now we're into 2002-3 and you begin to have a critical mass of homes
passed which are already penetrated. But by that time digital video will
have already begun to take hold in significant numbers on regular HFC
systems. So, does this make driving fiber to the home redundant? No.
HFC will support a number of digital channels of varying functionality and
utility, while eating into a limited amount of bandwidth. This fact stands out
starkly, due to the constant tension between program services which pay
the electric bill and rent and Internet access which many MSO managers
view simply as a must-carry proposition to round off the bundled service
mix. And then there is the 1 GHz bandwidth constraint that is imposed by
black coaxial cable.
The all-fiber alternative, on the other hand, will only be constrained by the
momentary higher costs of wdm during the early stages of deployment,
when we simply don't use it anyway. Even with the forbearance I've
mentioned exercised during early deployments, we still arrive the
advantages cited earlier, where Internet access is concerned.
As VCSEL costs come down with time, lambda counts will grow to
accommodate smaller and smaller clusters of users, with first, an n-fold
wdm-factor increase in baseband data rates, followed by yet another
order of magnitude in speed when 100 GbE deploys. I wont speculate on
multi-hundred gigabit or terabit Ethernet speeds at this time, much less
beyond.
You actually wind up with opposite dynamics taking place here, where
inhibit-enable factors for the two models are actually inverted with time,
when viewing the coaxial scheme which is limited to 1 GHz of bandwidth
and the ever increasing demands placed on it, against that of the fiber
model whose bandwidth is virtually unlimited, for our inability to tap it
entirely, at this time.
Now, let me look for that hard copy OSP magazine.
FAC |
