That was some excellent recall, Bernard. Although I wouldn't expect WTC to move over on the bench, just yet.
There was a time when I was in charge of all special services loops in Manhattan, as a Liaison to NY Tel from AT&T (hammer man du jour at the time) the ones that carried other than Bell System data services, that is. But that was eons ago. So, let's not get Tim too P.O.'ed here. We still need his input. [smile]
Dr. Loop, as George Hawley has been called, wrote the following article for Telephony Magazine back in '96, and it retains itself high on my list for its reference value. The article contains insights into the paper you alluded to.
internettelephony.com
It is best viewed at the site itself [click on the url] since there are some gifs included. I've posted the text below, for posterity.
Regards, Frank Coluccio
ps - Mike, this should keep you busy for a while, I should think ;-)
Note: There are other works by George Hawley that can be sniffed out at the Telephony site by searching the archives. I believe that there are also some links at the Diamond Lane Comms Corp site dlcc.com which, incidentally, has been acqired by Nokia.
=================
ADSL data: The next generation
ADSL is one of the leading contenders for providing
high-speed data access. Here's how the next generation ADSL
data access system could be improved
GEORGE T. HAWLEY
The buzz over asymmetrical digital subscriber line began when telephone
companies were experimenting with video dial tone services. Times have changed:
The Internet has become the "killer application," and attention has turned to the use
of ADSL for Internet data access. Current trial systems won't cut the mustard for
mass deployment of data services. Fortunately, a new generation ADSL data
access system has arrived in the nick of time.
Currently, the Internet is all the rage. The World Wide Web offers access to a
cornucopia of eclectic text, sound, image and video communication opportunities
with the click of a mouse. However, like a tantalizing treat just out of reach,
access to this panoply of information through dial-up voice modems is limited by
the 3000 Hz channel bandwidth.
A variety of improvements--including ISDN, ADSL, cable modems and several
wireless schemes--promise to improve the bandwidth situation.
Wireless, however, has its own problems. The current cellular alternatives at 800
MHz and 2 GHz are already resorting to digital technology to squeeze voice traffic
into less than half the data rate of a 28.8 kb/s voice modem. And wireless data
suffers from the same security issue as cellular telephony and cable modems:
eavesdropping.
ISDN has experienced a rebirth due to the pull of the Internet for more access
data throughput. At 128 kb/s, basic rate ISDN--as conceived more than 20 years
ago--is faster than voice modems but still too slow for high-speed image and video
transfer. Because ISDN is not a lifeline service, most ISDN users retain a second
line for POTS. And like voice modems, ISDN relies on dialed connections through
local telco voice switches.
The use of voice switches for flat-rate connections to Internet service providers
(ISPs) competes for dial tone with 911 calls and other important voice traffic.
That leaves just two realistic alternatives to sate the appetite of Internet users who
seek added bandwidth: cable modems and ADSL, pitting hybrid fiber/coax against
twisted pair copper.
Cable modem systems modulate digital data into a 6 MHz television channel slot
for multiple subscribers to share. With an example modulation scheme of 64
quadrature amplitude modulation (QAM), up to 30 Mb/s of raw data can be
carried downstream in a 6 MHz bandwidth, assuming a modest signal-to-noise
ratio in the channel.
But users are competing with one another for data access through the shared pipe,
and speeds can be significantly slower during peak usage periods.
The reverse channel is a greater challenge. After 25 years of promises of
two-way cable, only about 5% of the cable systems in the U.S. have invested in
the combination of hybrid fiber/coax technology and reverse channel upgrades
needed for reliable two-way data transmission.
Even with two-way operation in place, cable modems need to be very
sophisticated to work around the noisy 5 to 30 or 40 MHz spectrum set aside for
reverse channel communication. Because the modems can't sense the noise they
must combat, they need to wait for a central controller to tell them when to send
and at what frequency.
These issues can be addressed with new and rehabilitated cable plant and
sophisticated electronics--at a price. A cable operator must upgrade the entire
distribution network before the revenues can begin. That adds a risk that some
cable TV companies may not accept, especially with ADSL emerging as potential
competition.
The Right Performance
In the context of this article, ADSL means both symmetrical and asymmetrical
data signals for Internet access that share twisted pairs with POTS and that use
modern signal modulation techniques to accomplish the data communications task.
This is different from the complex American National Standards Institute
T1-413.1996 ADSL VDT standard (see sidebar).
In the more general context for Internet access, ADSL needs to accomplish only
three tasks: two-way data communications over twisted pair copper in frequency
bands above those dedicated to voice (for end user applications), two-way data
communications for operations support between customer premises equipment and
network equipment, and two-way communications between CPE and network
equipment to monitor and test the link.
For a given modulation scheme, ADSL has only a few key issues: the distance of
operation vs. data throughput, degree of asymmetry, transceiver power
requirements, transceiver size, transceiver agility and cost. There's no magic to
twisted pair transmission. Obvious impairments of attenuation and noise--especially
crosstalk from other pairs--need to be overcome. If one wants more bandwidth,
one must accept shorter distance operation for a given error performance.
The telephone loop falls into three basic categories: nonloaded loops up to 18,000
feet (about 70% of all loops), loaded loops longer than 18,000 feet, and
electronically derived loops. A typical electronically derived loop involves a digital
loop carrier system and a nonloaded loop beyond the DLC remote terminal (Figure
1).
Like basic rate ISDN, ADSL won't work through loaded loops. And although
ISDN is compatible with most derived loops, ADSL can't be supported realistically
by older generation, T-1 based DLC systems but by fiber systems.
Figure 2, derived from data supplied by AT&T Paradyne, illustrates one possible
trade-off between ADSL upstream and downstream data rates vs. 26-gauge loop
length. Telco planners would like to eventually offer up to 6 Mb/s downstream and
up to 640 kb/s upstream with an ADSL operating range of up to 18,000 feet (or
1300 ohms loop resistance, whichever is shorter). To get started, planners believe
that 1.5 Mb/s downstream and 384 kb/s upstream will be sufficient and will
support expansive Web surfing and real-time videoconferencing. These rates will
reach close to 18,000 feet, a pragmatic compromise aimed at the best combination
of performance and economy.
In addition, telcos would like some flexibility in data rates based on line conditions.
This would be comparable with the ability of voice modems to work at one of
several standard data rates, depending on channel characteristics.
ADSL technology by itself is a necessary but insufficient ingredient for mass
deployment of data services. ADSL transceivers must integrate into an access
system that respects 120 years of telephony evolution to minimize the total cost of
data access while maximizing the end user benefits. Unlike today's trial equipment
that might be characterized as "ADSL in a (very expensive) shoebox," the next
generation ADSL data access systems must blend their technology with a variety
of other key requirements to support mass deployment.
The next generation ADSL data access system needs to be designed to enable
telcos to aggregate all forms of data access traffic from DS-1 rates to optical
speeds. This will relieve telcos of having to install overlay access systems for each
category of data service: Internet, frame relay, ATM cell relay and local area
network emulation services. This target network requires that next generation
ADSL data access systems incorporate an ATM architecture similar to the
"service access multiplexer" described in Bellcore GR-2842.
ATM to the Desktop
Assuming the next generation ADSL data access system is an ATM cell
multiplexer, the greatest economy will be achieved in the long run by extending
ATM cells through the ADSL line to the end user, relieving the multiplexer of
having service-specific ADSL line cards with attendant adaptation and
segmentation/reassembly functions. By pushing these functions to the CPE, they
can be integrated through a standard interface into the PC and accomplish many
of the functions in software with the high-power microprocessors that are
dominating new PCs.
This approach opens the way for the PC applications to support multiple ATM
services in addition to Internet protocol over ATM. These benefits come at the
expense of some bandwidth dedicated to ATM cell overhead. Therefore, it is
logical to use ATM operations, administration and maintenance cells for operations
messages between CPE and the next generation data access multiplexer rather
than dedicate separate overhead channel bandwidth for an embedded operations
channel.
Bell regional holding companies are in the throes of evolving their operations
support systems (OSSs) from the function-specific TIRKS/
LFACS/COSMOS complex of Bellcore-supported embedded software to the
hierarchical Telecommunications Management Network (TMN) structure,
following the lead of the International Telecommunication Union. The element
manager is at the lowest level of this hierarchy as the equipment-specific software
interface to the rest of the network management software structure.
Figure 3 illustrates the role of the element management software as part of the
next generation ADSL data access system. Residing in a telco workstation, the
element management software must interwork with higher layers of network
management systems and even with embedded OSSs, while providing a graphical
user interface for intuitive use of the ADSL data access system features. The
element management system will ideally support simple network management
protocol with an evolutionary path to the common management interface protocol.
Also shown is a PC-based craft interface to facilitate system installation and
on-site diagnosis of potential faults. The multiplexer is shown with an Ethernet
10BaseT interface into the network management backbone. This high-speed LAN
interface will enable fast remote downloading of new software releases from a
central location.
As part of a total systems approach, it is essential for the next generation ADSL
data access system to provide for derived loops to serve customers more than
18,000 feet from the central office (or closer if appropriate). This requirement
takes two forms: one for the case of older generation DLCs and the other for
newer, next generation DLCs. In the case of an older DLC that may be served by
T-1 lines, an overlay addition is required to extend ADSL capability to the DLC
remote terminal site. This requires the next generation ADSL data access system
to extend its ADSL capabilities from the CO to a remote ADSL line card shelf,
preferably over optical fibers to accommodate data rates up to 6 Mb/s per line.
Depending on backplane design, a next generation Sonet-based DLC will have the
capacity to integrate ADSL into POTS line circuits to provide an integrated remote
interface that is much more efficient than the older DLCs, as might be expected.
The next generation ADSL data access system aggregates data in the CO from all
sources into a common Sonet ATM backbone facility. T
George T. Hawley is President of Diamond Lane Communications Corp.,
Petaluma, Calif.
ADSL primer
ADSL is currently available in two forms: carrierless amplitude/ phase modulation
and discrete multitone. CAP was developed by AT&T Paradyne, and DMT
variations are being developed by Amati, Aware/Analog Devices, Alcatel and
Orckit. When a standard was being considered for video dial tone by ANSI T-1,
DMT was selected on the expectation of early commercialization, although CAP
has reached the commercial market more quickly.
CAP, developed at AT&T Bell Laboratories in 1989, uses a variation of QAM to
create constellations of amplitude/phase pairs, each representing a group of bits.
CAP-16, as an example, transmits a constellation of 16 amplitude/phase pairs.
Data rates can be increased with CAP by using a bigger constellation and by
increasing the transmission rate of the amplitude/phase pairs.
DMT uses multicarrier modulation, first suggested by Burt Saltzberg and Steve
Weinstein at Bell Labs in 1968. DMT slices a frequency band into several
hundred--typically 256--sub-bands, each of which carries a signal modulated with
part of the data stream. The technique used to convert bits into frequencies for
transmission over twisted pairs is the Fast Fourier Transform algorithm. At the
other end of the line, the received signal is processed using the inverse transform
to recover the digital data. Data rates can be adjusted with DMT by increasing the
number of sub-bands and by the number of bits carried in each sub-band.
Although DMT was chosen as the basis of the T-1 standard, readiness for timely
and efficient commercialization makes CAP the leading contender for the
immediate needs of Internet access.
Powers of concentration
As the number of ADSL users increases, an access concentrator
can help provision and manage them more efficiently
BENJAMIN "TAC" BERRY
internettelephony.com
|
