| |
|
August 3, 1998
 ADSL trims down with
G.lite
Slow to satisfy its projected deployment
numbers, ADSL must now rely on the G.lite standard for consumer appeal. As computer and
telephone companies push G.lite, ADSL chip-set vendors are upgrading their products for
consumer-grade modems.
David Marsh, Contributing Technical Editor
Stephen Kempainen, Technical Editor
Wide-scale proliferation of ADSL (asymmetrical-digital-subscriber-line) access to the
Internet has stalled as a result of confusion over industrywide standards and unexpectedly
high implementation costs. In the United States--currently ADSL's prime territory--cable
modems have surged ahead in providing home access to high-bandwidth services. Now, the
Universal ADSL Working Group (UAWG) is rallying around the International Telecommunication
Union's G.lite specification for a lower cost, longer reaching, and splitterless ADSL
standard. By solidifying a global standard, IC vendors can optimise their products for
consumer-grade modems. But questions still remain over ADSL's widespread deployment.
Back in December 1996, proponents of ADSL deployment predicted 5 million connections by
2000, but recent, more realistic predictions suggest approximately 1.4 million. This
significant shortfall is principally due to the lack of standard, end-to-end network
protocols coupled with uncertainty over line-coding techniques. Furthermore, ADSL field
trials demonstrate that widespread deployment is more expensive than anticipated because
of spectrum-compatibility issues and anomalies in twisted-pair cable plant. But
discrete-multitone (DMT) line coding now has won the standardisation battle, and consensus
holds that asynchronous transfer mode (ATM) will provide networkwide compatibility.
Thus, chip-set designs reflect various levels of ADSL maturity--with new designs
featuring integrated ATM access.
Full-rate ADSL slims to G.lite
Full-rate ADSL transmits information to the consumer downstream at up to 8 Mbps,
accepting response data "upstream" at up to 800 kbps. These data rates depend
heavily on line conditions, including length. Full-rate ADSL needs a filter, or
"splitter," to separate the digital data from analogue telephony, permitting
both services to run concurrently. But the splitter requirement implies engineer visits to
customer premises, adding cost. By contrast, G.lite exchanges data at 1.5 Mbps downstream
and 512 kbps upstream (maximum) and needs no splitter. G.lite makes it easy for consumers
to install modems and increases ADSL's reach to virtually all households (Figure 1).
The UAWG's marketing boost and the emergence of the ADSL Forum's end-to-end
architecture restore ADSL's credibility. In the United States, telecomm companies are
currently deploying ADSL; trials are occurring throughout Europe, with target deployments
by year-end (see sidebar "Europe checks out ADSL").
Questions of whether a single worldwide standard is possible revolve around myriad
regional telephone systems that need accommodation. ADSL's promoters are rallying around
1.5-Mbps G.lite because they perceive cable-TV operators' broadband-access advantage. But
high-bandwidth telephony-based services such as multimedia gaming, video- conferencing,
and video-on-demand (VoD) still demand full-rate ADSL.
Optimising the cost-to-performance trade-off has ADSL-IC vendors offering products that
target varying bandwidth requirements (Table 1). Vendors base
their chip sets either on a general-purpose, high-performance DSP or on a lower
performance DSP coupled with coprocessor blocks. The high-perform-ance DSP route is more
flexible, but the DSP needs top-end performance to keep up with G.lite's real-time
processing requirements or ADSL's full-rate speeds. Alternatively, coprocessors can
perform parallel tasks, such as trellis coding and echo cancellation, and can
significantly reduce overall system power consumption.
Varied architectures make straightforward comparisons of multiple chip sets difficult.
Some vendors, such as Analog Devices, include all of the chips you need for a modem with
their chip sets--and for an inclusive price. Other chip sets may include the transceiver
and analogue front-end (AFE) in the price, but they exclude a controller and line driver.
To make a realistic comparison, ask vendors to supply a bill of materials for completing a
modem using various chip sets.
You'll notice that ADSL chip sets claim to comply with numerous versions of ADSL
standards. Don't assume that chips claiming compliance to the same standard will
necessarily work together. The problem is maturity--that is, vendors often claim
compliance with standards that are not yet final. The base-line ADSL standard is ANSI
T1.413 Issue 1, Category 1. Category 2 is a higher performance option that includes
trellis coding and echo cancellation for higher bit rates over longer distances. The
follow-on specification, T1.413 Issue 2, is now complete and awaits final ratification.
Issue 2 resolves ambiguities in Issue 1, and it adds a rate-adaptive operating mode and an
optional ATM interworking layer.
Vendors compete for territory
Currently, the only Issue 1, Category 2-compliant chip set available is Motorola's
CopperGold transceiver that centres on the company's MC145650 DMT transceiver IC. This
device integrates a general-purpose DSP, a T1.413-compliant data interface, and the AFE.
The embedded 563xx-series 24-bit DSP core processes 20-bit precision DMT algorithms;
coprocessors handle error-correction and echo-cancellation algorithms. Dedicated hardware
allows the DSP to work at lower clock speeds and saves power. (CopperGold's transceiver
typically dissipates less than 1W.) Motorola also offers the line driver, ATM cell
processor, and a choice of host-controller options to complete an ADSL modem. Developers
can also purchase an evaluation board, complete with application programming interface and
system analyser software.
Texas Instruments' ADSL chip sets provide three configurations--one for network-side
modems and two for client-side modems. TI's chip sets comply with ANSI T1.413 Issue 2,
Category 2 and support 8-Mbps downstream and 800-kbps upstream data rates. The TNETD2000C
chip set suits exchange equipment, with two-line support from one transceiver. Two
full-rate ADSL lines suit an exchange's DSL access multiplexer (DSLAM) as well as ADSL
line cards; both lines are programmable for upgrades as standards evolve or subscriber
bandwidth requirements change (Figure 2). TI's client-side
modem ICs comprise the TNETD2000R chip set that suits external- and network-termination
modems and the TNETD2000P chip set for PCI-card modems. The 2000P includes an ATM host
interface controller and Microsoft Windows drivers. A reference design will be available
by third quarter.
Alcatel, which pioneered ATM as the transport protocol for ADSL, offers silicon that
includes the Issue 2 option for ATM interworking. DynaMiTe is the company's
third-generation ADSL chip set in which the DMT modem IC integrates the ATM framing
function. The chip set doesn't include ATM's segmentation and reassembly (SAR) because
this function takes only approximately 2% of a host Pentium's bandwidth. But to remove any
real-time constraints from the host processor, DynaMiTe includes a dedicated controller.
Alcatel also provides a reference design for its chip set.
Polite local loop meets ATM
To promote rapid ADSL take-up, Alcatel licenced the DMT/ATM transceiver and AFE designs
to AMD, STMicroelectronics (formerly, SGS-Thomson), and Integrated Telecom Express, but
licensees must develop their own controller ICs. STMicroelectronics' Tosca chip set is the
first option to offer integrated ATM transport and framing functions. The Utopia level-2
interface connects directly to ATM systems. Using the vendor's mC code, the chip set can
operate as a remote ADSL terminal unit (ATU-R) or central-office (exchange) device
(ATU-C). A transmission "politeness" feature attenuates signal levels to reduce
crosstalk in short local loops. In the downstream direction, an external multiplexer
selects appropriate attenuation for reliable reception.
Analog Devices' second-generation AD20msp918 IC also adds Issue 2's ATM framing
options, complying with Issue 2 Category 1, as well as higher perform-ance trellis coding.
The company's designers chose to omit Category 2 echo cancellation, believing that reach
is more important than data rate. Crosstalk increases when using echo cancellation for
higher data rates, reducing effective reach. With trellis coding, the bit-error rate (BER)
remains acceptably low over longer distances. Suiting applications in which ISDN
(integrated-services digital network) shares the same twisted-pair line, the IC features
programmable frequency division between upstream and downstream spectra.
The latest ADSL chip set introductions from Lucent and Rockwell sidestep ANSI
standards, bidding for consumer acceptance with a combination analogue-and-ADSL modem.
Lucent's WildWire chip set evolves from its ITU V.90 (56-kbps) offering by adding G.lite
capability (Figure 3). The DSP1690 IC couples two DSP1600
cores in one chip to provide 200-MIPS processing capability, typically consuming 1.5W. The
chip set includes an autodetect feature that determines if the loop provides a DSL
connection, allowing the user's modem to exchange data at the highest available rate.
Software upgrades to G.lite compliance will become available when the standard is
complete. Details for Rockwell's chip set were just emerging at press time, so check with
the company (www.rss.rockwell.com) for full
details.
Testing ensures success
With the diverse array of ADSL chip sets available, interoperability testing is
critical to success. Chip vendors must track evolving standards and cooperate with one
another to guarantee physical- layer compatibility, while modem vendors need to cooperate
at the network- protocol level. The ADSL Forum steers these issues, contracting
independent test labs to develop compliance test specifications for ITU and ANSI
standards. Individual companies are also taking action. For example, Alcatel produces an
open implementation guide that describes the choices for the company's chip set in the
standards context, notably with respect to ATM layer choices. And Alcatel, Analog Devices,
and Texas Instruments have jointly announced intentions to perform mutual interoperability
testing.
Universal interoperability testing should complete the UAWG's consumer-grade modem goal
and restore ADSL to the forefront of the race for the home information superhighway.
Because IC vendors are newly confident that ADSL will fulfil its initial promise, you can
expect more highly integrated and application-specific chip sets to appear soon.
Acknowledgments
Thanks to Rupert Baines at Analog Devices and Jean-Claude Baumer at Texas Instruments
for assistance in preparing this article.
Congratulations. You've picked a programmable-logic vendor, an architecture, and a
target device. Now, how do you ensure that you'll achieve a high degree of logic
usage--not just with those test circuits you used for the evaluation stage, but with your
real design?
First, rely on your vendor's knowledge of its devices. Your expertise is with the
system, not with every feature of each chip within it. Instead of diving into the design
headfirst, spend a few days, and really read the data sheet, appropriate application
notes, and user manual for both front- and back-end design software. You'll no doubt learn
a few tricks that will measurably improve your results.
Second, use predesigned circuits if they make sense. Although using a vendor's
macrofunctions and cores may feel like a blow to your talented engineering ego and
instantiation makes your design less portable to other vendors' programmable-logic devices
and ASICs, the people who created the macrofunctions understand their chips a lot better
than you do. They can probably create a circuit that is more reliable, runs faster, uses
fewer gates, and burns less power than yours.
Third, learn from the expertise of your peers, both at your company and elsewhere, who
have used the device. Internet newsgroup comp.arch.fpga is a tremendous source of useful
information and--usually--intelligent exchange of ideas on a variety of programmable-logic
topics. I am amazed not only by how many vendor representatives participate in the
discussions, but also by how many consultants log in and how much good, free advice they
share.
Speaking of consultants, my fourth recommendation is to consider tapping into the
knowledge base of one of these individuals and to do it before the 11th hour. Using an
outside expert can be cheaper than attempting the design in-house. This situation is
especially true when you consider the tool acquisition, the learning curve to get to
enough proficiency to drive the tools, and the iterations to get the design right.
Consultants use a variety of devices from multiple manufacturers and have seen lots of
designs, including the most challenging ones you can imagine.
A little money invested up-front on device and design advice may reap significant
returns down the road. However, insist on a confidentiality clause in the contract, and
make sure the consultant provides all the necessary files and clearly documents the
design, so that you can understand and maintain it in the future.
Finally, don't work too hard. Over the last several years, programmable-logic devices
have significantly increased in logic capacity and decreased in cost per gate; beyond a
certain point, your continued efforts at squeezing more logic into a part reap diminishing
returns. Bite the bullet, switch to a bigger device, wrap up your design, and get on with
your life. Just make sure you can still hit your timing with the larger device. One
secondary benefit of this approach is that if marketing comes in with last-minute
feature-set changes that increase design complexity, you'll have the gate-count head room
to ensure performance and pin locking.
With all of the media attention that asymmetrical digital-subscriber line (ADSL) is
receiving recently, you may be lulled into thinking that your Internet bandwidth problems
are just a phone call away. But, realistically, what sort of services should you expect in
the near future, and how do Europe's widespread integrated-services digital networks
(ISDNs) fit into the picture?
Europe's two largest telecomm operators are Deutsche Telekom (DTAG), having 45 million
subscribers, and British Telecom (BT), which has 28 million customers. Both companies have
invested heavily in ISDN and have substantial interests in ADSL. BT originally proposed
ADSL as the enabling technology for video-on-demand (VoD), so the company ran ADSL-based
trials with approximately 2500 test consumers from 1994 to 1995. But on its own, VoD
failed as a commercial proposition. ADSL's emerging commercial service model adds Internet
access, multimedia, online services, and video- conferencing for business and residential
users. The Internet- access requirement is huge and growing fast: DTAG's T-Online,
Europe's largest Internet-service provider, currently handles 60 million calls per month
and has reported 60% growth over the company's 1996 and 1997 earnings.
ADSL must live with ISDN
Compatibility with ISDN is critical for ADSL's success in Europe (see sidebar "ADSL complements ISDN"). ISDN is a huge success in Germany,
where DTAG's Hans Gusbeth confirms that the company has 3.3 million basic-rate lines and
more than 8.5 million channels in use. In the United Kingdom, BT declines to confirm how
many ISDN lines the company leases, but independent 1996 figures suggest 750,000 lines,
and BT's current advertising boasts 2000 new installations per week.
ISDN's design doesn't accommodate simultaneous digital and conventional analogue
telephony over the same line, but that isn't a problem. Digital phones guarantee the
integrity of the voice circuit over ISDN because they can take power from the incoming
phone line, ensuring life-line service. But with so many subscribers having only ISDN
connections, ADSL must be simultaneously deliverable with ISDN. Because ADSL's original
specification conflicted with ISDN, DTAG initially rejected the new technology, but soon
the company became a prime mover in redesigning the specification. At the CeBIT '98 show,
held in March in Hannover, Germany, DTAG's Gerd Tenzer affirmed his company's commitment
to xDSL technologies--and ADSL in particular.
Prepare for ADSL roll-outs
BT's and DTAG's technical trials are now complete, and both companies are busy
test-marketing ADSL-based applications. DTAG started pilot projects during April to
approximately 100 commercial and 300 business customers, as well as to a university
campus. Service-provider partners will assess subscriber acceptance of broadband
services--including a business TV channel and an animated multimedia shopping
mall--storing source material directly on DTAG's servers. Building on this experience,
DTAG expects to roll out xDSL systems in eight major German cities before the end of this
year: Berlin, Bonn, Cologne, Dusseldorf, Frankfurt, Hamburg, Munich, and Stuttgart.
Furthermore, the company plans to have 40 xDSL networks operating next year--with 70 in
place by 2002.
BT plans to provide ADSL links to approximately 2500 business and residential customers
in West London this summer. BT's new trials involve approximately 30 independent service
providers who will furnish Internet access, online services, and VoD. The downlink will be
2 Mbps with 148 kbps upstream and will be based on standard discrete-multitone (DMT)
technology. According to David Orr at BT's press office, the organisation is fully
committed to ADSL and will roll out full-blown services in "suitable locations,"
provided the West London trials succeed.
Other European trials include Telecom Italia's Milan venture, in conjunction with
AMUSE, the European Commission's Advanced Multimedia Services to Residential Users
programme. Technical trials began last year with an 8.2-Mbps downlink and 640-kbps uplink
that suits VoD delivery using MPEG-2 video streams. Similarly, Belgium's Belgacom, France
Telecom, Norway's Telenor, and Kingston Communications (a UK independent) are all offering
video-bandwidth services on a trial basis. Operators in Finland, Ireland, the Netherlands,
Spain, Sweden, and Switzerland have trials that target ADSL's contemporary mixed-service
commercial model.
Where will ADSL reach?
BT's "suitable-locations" phrase masks myriad commercial and technical
considerations that vary from country to country. The Scandinavian countries enjoy the
largest installed base of PCs in the world, with Sweden boasting the largest number of
telephone lines per capita--68.3 lines per 100 people compared with 60.2 lines per 100
residents in the United States. (Note that those are 1994 figures.) In the Netherlands,
cable companies have achieved close to 100% penetration, so coaxial solutions might look
more attractive than ADSL. But the weak Italian cable infrastructure drives Telecom
Italia's aggressive ADSL roll-out plans, which are scheduled to begin next year. Other
operators with major ISDN investments want to protect their interests but may come under
separate attack from the European Union's 1998 telecomm deregulation activity.
What's more quantifiable is the number of households that ADSL can potentially reach.
Italy has the shortest average local-loop lengths anywhere, with almost all subscribers
within a 3.7-km carrier-serving-area (CSA) range and 6.1-Mbps delivery. In the United
Kingdom, more than 85% of subscribers are within CSA range, and in Germany that figure
reaches 80%. Virtually all European subscribers fall within 5.5-km and 1.5-Mbps delivery
range, which is also the specified reach for basic-rate-interface (BRI) ISDN lines.
Moreover, wiring practices differ from nation to nation. Some countries, such as Italy,
use "bridge taps," which are unterminated spare pairs on the line that can
resonate unpredictably. In some regions--including the United Kingdom--not all lines are
copper. Aluminium-based connections are approximately 50% more resistive than copper per
unit length, reducing reach. What's worse, the impedance change where copper meets
aluminium creates reflections. Wire gauge is also significant, and, again, impedance
changes between different connections affect performance. Although more of a problem in
the United States, very long lines may have inductive load coils that maintain
high-frequency voice response but kill ADSL-frequency signals. These considerations mean
that operators must survey their lines to ensure that full-speed ADSL links are viable.
BT is considering test-marketing G.lite, which can theoretically reach almost every UK
household. Implementation is a challenge, however, because the company has no control over
the end user's intended location and therefore has no qualification of local-loop quality.
But in February, BT became the first European telecomm operator to join the Universal ADSL
Working Group and is now working to solve practical implementation issues. Because this
process is so new, the company has no timetable for G.lite roll-outs.
What about the backbone?
You might wonder where the backbone bandwidth will come from to support the mass of
new, potentially multi-Mbps subscribers. According to Robert Bury, marketing manager at
Alcatel Microelectronics in Belgium, "ATM is emerging as the clear transport winner
for ADSL applications. ATM's bit-rate flexibility matches the ADSL/DMT rate-adaptive model
extremely well. Alcatel promoted this model from the start, initially with better
acceptance in Europe than in the US where the frame-relay model was more prevalent. But in
the last year, ATM has become the main model in the US, too." Bury's statement
explains why you're beginning to see ADSL chip sets with integrated ATM ports.
Meanwhile, the backbone grows stronger with techniques such as ultradense wave-division
multiplexing from Lucent Technologies. Lucent's WaveStar OLS 400G is a regenerative system
for optical fibre that can transmit 3.2 Tbps using eight 400-Gbps parallel connections.
Lucent estimates that WaveStar will work with approximately 90% of all installed fibre,
which would provide 80 times more bandwidth than you currently enjoy.
To understand integrated-services digital-network (ISDN) and
asymmetrical-digital-subscriber-line (ADSL) interoperability issues, consider the basic
technologies. ISDN's designers fitted it seamlessly within the increasingly digital
n364-kbps telephone system, and there's little difference between a normal analogue phone
line and an ISDN connection. For ISDN you need different wall-socket components to
terminate and split the digital transmission line, but the underlying communications
infrastructure is essentially the same. The big change for ISDN is that the codecs that
translate between voice and the phone system's native PCM format move from your local
exchange into your phone. If you're just transmitting data, you can benefit from an
all-digital connection into your local trunk.
An ISDN basic-rate interface (BRI) provides two 64-kbps voice/data channels and one
16-kbps signalling/control channel, respectively termed "2B+D." If your
equipment permits, you can concatenate both data channels for a 128-kbps data connection.
Germany uses the "4B/3T" pulse-amplitude-modulation (PAM) line-coding technique
to encode the data; the rest of Europe (and the United States) use "2B/1Q"
coding. Either coding system relies on combining two or more bits into one symbol.
"2B/1Q" means two-binary, one-quaternary, signifying coding 2 bits into one
four-level element called a "quat." An ISDN signal's frequency spectrum depends
on the line-coding technique but contains energy up to 110 to 120 kHz for the higher
bandwidth 4B/3T code.
Most current ADSL implementations in the United States use carrier-amplitude-phase
(CAP) modulation, a more complex version of an analogue modem's
quadrature-amplitude-modulation system. CAP's competitor is discrete multitone (DMT),
which communicates using a number of discrete frequencies, or "bins." DMT has
won the standardisation battle that's now embodied within the baseline ANSI specification
T1.413, and CAP is not part of any standard. But this move did nothing to ensure that ADSL
could coexist with ISDN.
In the original specification, the DMT upstream spectrum comprised 32 tones spaced
4.3125 kHz apart, with tones one to five and 32 remaining unused. Low-frequency usage
extended from approximately 25 kHz, which was well above analogue phones at 4 kHz but well
within ISDN bandwidth (Figure A). For ADSL to coexist with
ISDN, the ADSL data link must use higher frequencies than originally envisaged. Solutions
include shifting tones up using a mixer, as well as extending the FFT algorithm to
accommodate higher frequencies (Figure B). Consensus
recommends shifting the base ADSL spectrum up from ISDN, with a corner frequency around
130 kHz to separate both systems. The newly extended ADSL upstream spectrum comprises 64
tones with the same 4.3125-kHz frequency spacing, but with tones one to 32 and 64
remaining unused (Figure C). |
