The State of ADSL #1
The Bandwidth Played On
While BT continues to charge an arm and a leg for
its ISDN services, Steve Gold takes a look at a
comms technology that will make ISDN look as slow
as a 300bps modem
Although it hardly seems possible, V.34 modem speeds - 28.8Kbps
and, latterly, 33.6Kbps - have been commonplace in the comms
industry only for 18 months or so. Just a few years ago, the highest
speed modem users could eke out of the PSTN was 14.4Kbps. That
was no doubt why BT priced its ISDN-2 service, which offers speeds
of 64Kbps, at just under œ400 for installation and roughly as much
again for an annual subscription.
Despite the fact that these rates are almost eight times what state
monopoly Deutsche Telekom charges its ISDN subscribers in
Germany, BT has persisted in offering ISDN as a value-added
telecoms solution to its customers.
Some experts have suggested Deutsche Telekom and, to a large
extent, France Telecom, have been able to offer ISDN at cut-price
rates because of their virtual monopoly on inland calls in their
respective countries.
But even BT's senior ISDN management dismisses such arguments,
claiming that ISDN has required a significant investment in bandwidth
on the UK network.
Even this notion is patent nonsense, as most telephone lines no longer
connect to the phone exchange. In fact, they terminate at a device
known as a line concentrator just a few hundred yards from offices
and houses in city areas, and in clusters in the countryside.
These devices concentrate the local loop voice signals on to one or
more 2Mbps data channels feeding into the telephone switch. The
concentrators also transfer the inbound analogue audio channel into a
digital data feed, using a codec.
Codecs can also be found in digital mobile phones. More powerful
than a 100MHz Pentium processor, they cost about œ100 each. But
their functionality outweighs their cost, and BT has splashed out on
installing digital concentrators in most urban and city networks across
the UK.
Concentrators also allow BT to service a cluster of, say, 200 phone
lines with about 40 call channels into the exchange. The savings on
local loop connections offset the cost of the concentrators and their
associated codec technology.
BT would like to harness its powerful multimegabit network by
broadcasting digital multichannel TV signals. As Sky's progressive
dominance of the UK TV industry shows, broadcasting is where the
money is. But the government won't allow BT to start broadcasting
down its digital network, for fear that the company might go even
further than Rupert Murdoch has in broadcasting.
So BT has been exploring new technologies, one of which is
asymmetric digital subscriber line (ADSL).
ADSL allows data to be moved in one direction at up to 7Mbps, and
640Kbps in the reverse direction. The data flow can be reversed
several times a second, which means that ADSL's primary data flow
can take place in the direction of the main data stream from the
subscriber or the host system to which they are connected.
ADSL, like ISDN, has the great advantage of being compatible with
the copper twisted-pair cabling that almost all phone subscribers rely
on for the final leg of their links from the switch (or concentrator) to
their BT or cable phone socket on the wall. For a payment of about
œ400, providing the telecoms network supports ADSL, a phone
subscriber can plug an ADSL device straight into the phone line.
ADSL has caught on big time among the US telecos. In return for
roughly the same investment as ISDN, telcos can install ADSL
modems instead. Seven megabits is a lot of bandwidth. Put simply,
the networking bandwidth is in the same realm as ATM except, of
course, it is available right to the home or office.
Although BT is forbidden to broadcast in the current regulatory
environment, what would happen if it narrowcasts signals to
subscribers? Narrow-casting involves feeding a unique TV signal to
one subscriber and ADSL is the ideal environment for this - hence the
phenomenal worldwide interest in the technology.
BT is pilot testing ADSL on its UK network in a selected number of
cities.
The trials have proved that the technology works well. BT remains
tight-lipped on the subject of its ADSL plans for fear of alerting the
cable TV/phone companies to its actions, but sources suggest that it
will embrace ADSL as a narrowcast and datacoms medium par
excellence over the next six to 12 months.
ADSL offers almost unbelievable benefits to the network user. Seven
megabits will support a full bandwidth company Lan, although few
organisations could generate Wan traffic at that bandwidth. It also
offers the ability to multiplex several dozen voice channels together, as
well as a video conferencing link or two. In short, ADSL is a
networking nirvana.
BT is not unaware of this, but its primary focus is on narrowcasting.
Once the network is installed, few doubt that BT will capitalise on its
broadband links into the home or office. It has a near monopoly in
terms of the local loop outside those areas serviced by cable TV and
phone companies, and would have to be extremely stupid not to
capitalise on its resources.
In the US, ADSL is now being rolled out as a cable modem
technology to tens of thousands of homes and offices. It will not be
long before the user community realises ADSL's potential, and then
things will really start to motor.
ADSL has also started to attract interest from other kinds of
companies in the US. Early last month, Ameritech and IBM
announced they were teaming up in the Chicago area to test ADSL
links to homes and offices. The trial will involve some 200 customers
of both firms, concentrated in a part of the Chicago area which has
yet to be chosen. IBM says the plan is to offer channels of 1.5Mbps
each to subscribers to allows them to access the Internet.
By limiting the data speeds, IBM claims that standard voice calls can
also be carried on the same circuit, at the same time as the data
channel.
Ameritech says it is using prototype ADSL modems from several
manufacturers in the Chicago trials. The price of these modems is
about $1,500, although officials say that the production price will fall
to about $700.
Other trials announced so far include one launched in February by
GTE in Texas; a joint effort by GTE, Microsoft, and the University of
Washington announced in August; and a trial by Bell Canada which
started in two Canadian communities in September.
As part of the Chicago trials, IBM will provide Internet and intranet
services, as well as various network and datacoms resources through
its IBM Global Network operation.
Although it's still early days in the Chicago trial, IBM says it plans to
use the trial as a means of showing what ADSL has to offer the home
and business user, and, of course, that its existing network
infrastructure can handle the high data speeds involved.
The GTE/Microsoft trial in Redmond, Washington, meanwhile,
started in August of this year and will run until February 1997. The
Redmond ADSL trials held in conjunction with the University of
Washington centre on 40 Microsoft and GTE employees using ADSL
technology to test high-speed access to the Internet and private data
networks using Microsoft Windows NT-based servers and GTE's first
generation of ADSL modems.
Phase two of the test will increase ADSL coverage to more Microsoft
and GTE employees, along with the University of Washington and up
to 60 businesses in the Redmond area. This phase, which starts at the
end of this year, will include different types of ADSL modems, and
will also incorporate other Microsoft products like email, news
services, conferencing and electronic commerce.
GTE claims that, as a result of the trials, ADSL will be adopted on a
wide scale across several regions in the US. As a result, GTE predicts
that an ADSL modem will fall in price to about $250 by the end of
1997.
So where does ATM fit in with ADSL? As far back as April 1994,
Newbridge Networks announced that it was developing a hybrid
version of ATM which would work over ADSL modems. Back then,
of course, ADSL modems cost well into five figures and were only
found in the laboratory. Two-and-a-half years down the road, ADSL
is being used as a communications medium in the real world, and
several vendors of ADSL technology already support ATM as an
option on their units.
But it's not all wine and roses for ADSL, as several networking
companies have discovered that ADSL can't be provisioned on copper
circuits equipped with load coils or bridge taps. Furthermore, copper
cables more than 400 yards long cannot support the full 7Mbps data
bandwidth. This is one of the reasons why several of the US telcos
have limited their data speeds across ADSL links to 1.5Mbps, since
cables of up to four miles long can easily accommodate such
bandwidth.
ADSL is very distance-sensitive. With a copper loop of about
12,000ft, ADSL claims to support data streaming at up to 7Mbps. But
at more than 18,000ft, even 1.5Mbps ADSL signals can break down.
According to Ameritech, despite the fact that at least 80 per cent of
businesses and residences in highly populated areas of the US fall
within these thresholds, much of the wiring is old, cracked and wet.
While the copper local loop was designed to support an audio
bandwidth of about 6KHz, ADSL makes use of the full radio
spectrum up to 1.1MHz.
One investor, Bellcore, claims that ADSL has been designed to
accommodate difficult telecoms media, stepping down automatically
to slower speeds until the signal gets through. David Waring, director
of broadband local access and premises networks with Bellcore's New
Jersey operation, says line quality should not be a significant problem
if the telcos limit ADSL usage to a maximum cable run of 18,000ft -
significantly greater than most runs BT employs on its UK network.
A possible solution for far-flung telephone users is for the telco to
install a remote device known as a digital loop carrier (DLC). Copper
lines can connect to the DLC, which in turn can connect to a switch
via a digital circuit. As long as the subscriber is within 18,000ft of the
DLC, even the most remote of country satellite exchange links can
support ADSL to at least 1.5Mbps.
Perhaps more importantly, the ADSL modem manufacturers are
already wise to the problem of distance limitations and are already
working on a second generation of ADSL modems known as
rate-adaptive ADSL, or RADSL. These devices, which are expected
to ship in the second quarter of 1997, automatically check what sort
of ADSL speeds a copper cable can support and negotiate a link
accordingly.
According to Bellcore, RADSL modems can support step-down
speeds as low as 640Kbps in both directions at distances of up to
30,000ft - way above the normal limits for standard telecoms
operation - yet still supporting data rates 10 times that of ISDN.
Against this backdrop in the US, BT may face something of an uphill
struggle to continue marketing ISDN at its current price much beyond
the end of 1997. But before then, networking resellers can expect
some sexy new ADSL product launches from the computing and
networking majors, which are also working feverishly behind the
scenes on testing the new technology.
20-NOV-96 |
