IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 9. NO. 6. AUGUST 1991
`
`765
`
`High Bit Rate Digital Subscriber Line: A Copper
`Bridge to the Network of the Future
`
`Wil Walkoe and Thomas J. J. Stan, Member, IEEE
`
`Abstract-The high-bit-rate digital subscriber line (HDSL),
`permitting low-cost 1.5 Mb/s copper access, will ease the tran-
`sition to fiber access by accelerating the use of higher speed
`services. Copper will dominate over fiber customer access for
`at least the next ten years. During this period, the success of
`high-speed switched services will depend on the connectivity
`provided by both fiber and copper access. HDSL will initially
`be used to serve private-line DS1, ISDN primary rate access,
`and digital loop carrier feeders. Later, the HDSL will be ap-
`plied to switched services such as metropolitan area networks
`(MAN’S) and circuit switched DSl’s.
`
`I. INTRODUCTION
`
`F switching centers, is now widely endorsed as the
`
`IBER optics, long popular for transmission between
`
`transmission technology of choice between the switching
`center and the customer site. The fiber optic technology
`being developed now will be the most economical means
`to build new access facilities to customers using plain old
`telephone service (POTS). More importantly, fiber optics
`are essential for the success of broadband telecommuni-
`cations services. Telephone companies are rapidly de-
`ploying fiber optic cables in the access plant today. But,
`the enormity of the existing local access network will
`make the fiber optic facilities a minority of the local ac-
`cess for many years. The powerful economic pressures
`for reducing transmission costs and providing broadband
`services raise serious concems about the next twenty years
`required to fully deploy fiber. We suggest that while max-
`imum use is made of the fiber as it is deployed, we must
`also make the best use of the existing $100 billion copper
`cable plant in the U.S. The high-bit-rate digital subscriber
`line (HDSL), permitting low-cost 1.5 Mb/s copper ac-
`cess, will ease the transition to fiber access by accelerat-
`ing the use of higher speed services.
`
`11. AN ILLUSTRATION: PHOTOWN
`
`The upscale suburban community of Photown typifies
`the communications revolution of the mid-to-late 1990’s.
`The same fiber-to-the-curb network that provides POTS
`to Photown’s homes and businesses carries a powerful set
`of advanced services at a modest incremental cost. The
`smallest business sites have access to switched computer
`communications at throughput rates that were once pos-
`sible on only the largest private networks. Medical clinics
`access multimegabit x-ray and CT images from remote
`hospital record centers in a matter of seconds, and a va-
`riety of other bitmapped and coded image formats are rev-
`olutionizing applications ranging from computer-inte-
`grated manufacturing to the preparation of advertising
`copy. Even routine business paperwork is being auto-
`mated with computer-aided workflow systems based on
`scanned business documents.
`Local hotels have equipped many of their meeting
`rooms to support the growing demand for video confer-
`encing, and medium to large business sites typically have
`video facilities of their own. Many video services for
`home and business customers require full video transmis-
`sion only “downstream,” to the user, with simpler inter-
`actions using voice and low-speed data and control sig-
`nals providing all of the needed
`interaction
`in the
`“upstream” direction. These applications, including
`training and education services, library access to multi-
`media materials, and a growing number of public ser-
`vices, are deployed even more widely than videoconfer-
`ence facilities, penetrating individual offices and many
`residences. A growing body of material is being devel-
`oped for these applications on optical storage media, ac-
`cessible over the network at disk-transfer speeds.
`Photown is growing rapidly, as the needs of a growing
`service economy create strong demand for a modem com-
`munications infrastructure. Despite the virtually unpre-
`dictable growth pattern created by the work-at-home phe-
`nomenon, the Photown Telephone Company has little
`difficulty adjusting to new service demand wherever it
`springs up. Any line can be upgraded to advanced service
`
`Manuscript received October 19, 1990; revised May 1, 1991.
`The authors are with Ameritech Services, Rolling Meadows, IL 60008
`IEEE Log Number 9101626.
`
`0733-8716/91/0800-0765$01.00 0 1991 IEEE
`
`

`
`7 66
`
`IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 9. NO. 6. AUGUST 1991
`
`capability with the addition of the right terminal equip-
`ment at the subscriber’s site and a high-speed connection
`at the local exchange office.
`
`111. WHAT’S WRONG WITH THIS PICTURE?
`For an imaginary location, Photown has had a lot of
`visitors. Minor variations on the scenes described above
`appear regularly in network-of-the-future articles in the
`communications trade press, timestamped anywhere from
`the mid-1990’s to the turn of the century.
`Forecasts of this sort are based-in part-on valid tech-
`nical information. In volume production, known designs
`for fiber-to-the curb telephone systems could be cheaper
`than traditional copper telephone loops for most new con-
`struction in the 1993-1995 timeframe. These systems
`could deliver all of Photown’s advanced services at rea-
`sonable cost, as well as providing an excellent platform
`for the evolution of even more powerful applications in
`the early 21st Century. The only problem is that there
`might not be anybody to talk to at the other end of the
`high-speed call.
`Fiber optics will replace copper in the local telephone
`loop, but not all at once. The construction budget of a
`local exchange carrier (LEC) can support only a modest
`deployment of new cable plant each year. A typical LEC,
`devoting 100% of its new-construction budget to install-
`ing fiber, would take about twenty years to reach 50% of
`its customers with fiber-optic local service. Even if all
`rehabilitation projects were added to the fiber deploy-
`ment, it would still take nearly twelve years to reach 50%
`fiber penetration. New service opportunities could accel-
`erate this process, but there would still be a long delay
`between the first fiber deployment and widespread avail-
`ability of high-speed services. Fig. 1 illustrates the length
`of time it will take to replace copper transmission with
`fiber optic systems. Though fiber optic cable can be eco-
`nomically proved-in for POTS alone, the most exciting
`potential lies in the provision of higher speed services.
`The success of these services depends on connectivity
`within a large community of interest.
`The first fiber optic communities-the Photowns of the
`world-will be islands of high-speed services. Photown
`residents could easily be equipped for wideband commu-
`nication of computer data, digital images, and interactive
`video to one another, but calling would initially be con-
`fined to a small number of sophisticated customers. The
`resulting market for wideband information services could
`hinder the attraction of service providers. Photown will
`have high-quality telephone service and a cost-effective
`implementation of narrowband enhancements such as
`ISDN, but wideband “network of the future” services will
`have to wait for the rest of the world to come on line’.
`
`80%
`
`-
`70% -
`60% -
`50% -
`40% -
`30% -
`20% -
`10% -
`
`1980
`
`1985
`
`1990
`
`199s
`
`2000
`
`ZOOS
`
`2010
`
`2015
`
`2020
`
`2025
`
`Fig. I . Conceptual model of fiber deployment
`
`IV. THE MISSING LINK
`What would it take to bring wideband services to the
`world at large? The ideal solution would be a low-cost
`transmission system that could instantly convert any ex-
`isting local telephone connection to a wideband channel.
`No changes would be needed in the outside plant; the line
`termination equipment would simply be upgraded at both
`ends of the loop, with relatively simple electronics. If this
`could be accomplished with a line termination of about
`two to three times the complexity of a basic-rate ISDN
`NT 1, advanced services could be affordably deployed
`wherever a significant business opportunity exists-inde-
`pendent of geography.
`The service would not need to operate at SONET
`the services described in the
`broadband speeds. All of
`Photown fable could be implemented over digital lines ut
`the DSl datu rute of 1.544 Mb/s. However, the current
`T1 carrier technology for delivering DS1 service would
`not meet the needs of a new mass communication service
`because of its high cost, special requirements on the sub-
`scriber’s inside wiring, and long, complicated network
`provisioning process. For low-cost and timely provision
`of service, there is a need for existing telephone lines to
`transport DS 1 data rates without regard to repeater place-
`ment, cable-pair placement in a particular cable binder
`group, or its configuration of bridged tap wiring.
`We are closer to that solution than is commonly be-
`lieved. In November 1989, Ameritech conducted experi-
`ments together with Bellcore and one of our major equip-
`ment suppliers, in which prototypes of a new transmission
`technology known as HDSL (high-bit-rate digital sub-
`scriber line) successfully carried DS 1 signals over carrier-
`serving area distances on existing Illinois Bell telephone
`loops. The loops were not conditioned in any way, and
`they included some transmission challenges, such as
`bridged taps, that have caused problems with other so-
`
`‘It would be an exaggeration to say that there would be rio application
`of wideband communications, but the applications would be limited to spe-
`
`cial services within a narrow community of interest until fiber deployment
`reaches a critical mass. Based on Fig. 1, this process could take one to two
`decades.
`
`

`
`WALKOE AND STARR: HIGH BIT RATE DSL
`
`761
`
`phisticated signal formats. The HDSL system provided
`excellent transmission over these facilities, and an appli-
`cation demonstration delivered a high-quality videocon-
`ference signal over a distance of 12 000 feet to the serving
`exchange office.
`HDSL is an outgrowth of transmission techniques that
`were developed for ISDN basic rate access and high-speed
`digital modems. While several different detailed imple-
`mentations of HDSL are currently under consideration2,
`they all share the following basic characteristics.
`Transmission over existing distribution wire pairs.
`Full-duplex, DS 1 -framed transport at 1.544 Mb /s.
`Specified bit error rate ten times better than T1 car-
`
`rier.
`
`Repeaterless operation for full camer serving area
`(CSA)3.
`Operation in the presence of bridged taps and gauge
`changes.
`Spectral compatibility to permit coexistence in the
`same cable-sheath with voice, basic rate ISDN, T1 car-
`rier, and many other transmission systems.
`Today, HDSL is still a laboratory system, but it could be
`put into full-scale deployment in about two years. Some
`hard work remains to be done in the negotiation of trans-
`mission specifications and the development of operations
`plans for full-scale deployment. Yet, the feasibility of
`HDSL is well established.
`Upgrading a copper phone line to HDSL will not be
`quite as inexpensive as providing DS1 service via fiber to
`people in Photown, but the cost and provisioning of HDSL
`will make it attractive for most of the services described
`in the fable. And since most potential customers will be
`within reach of HDSL on the day it is introduced, mass-
`market wideband services will not have to wait for large-
`scale fiber deployment.
`HDSL deployment can lead to the emergence of wide-
`band service demand, because the technology can be
`proven-in with benefits to current services: private-line
`DS 1, ISDN primary rate access, and digital loop carrier
`feeders.
`
`V. WHAT ABOUT SWITCHING?
`HDSL can connect a mass market to the nearest tele-
`phone exchange at wideband data rates, but where will
`the signals go from there? Most of the services described
`in the opening fable will require wideband switched con-
`nections, not just cheaper DS1 private lines.
`Fortunately, a complementary technology is likely to
`become available in the same mid-1990’s timeframe as
`
`’TIE1 proposals for HDSL implementation include: one-pair QAM
`transmission with trellis coding (AT&T), two-pair 2B1Q (separate propos-
`als by Bellcore and BNR), two-pair multitone transmission (J. Cioffi,
`Telebit), and combined coding-precoding (MotorolaiCodex).
`’The CSA permits loop lengths up to 12 Kft of 24 AWG cable, or 9 Kft
`of 26 AWG cable.
`
`HDSL. Metropolitan area network (MAN) systems, ini-
`tially developed to provide a limited number of connec-
`tions at DS3 and higher speeds for major sites in down-
`town areas, have proven to be even more effective as DS1
`data switches for a larger subscriber base. MAN’S provide
`an excellent switching platform for DS1 data communi-
`cations, digital image transmission, and real-time inter-
`active video. The main limitation that has been antici-
`pated for these services is the small number of subscribers
`with DS1 access to the exchange-and with HDSL, that
`number can grow rapidly.
`A second wideband switching technology is also near
`readiness: circuit switching of DSl’s and fractions of
`D S l ’ s that function much like and ISDN B-channel call,
`but at a 1.544 Mb rate.
`The residents of Photown may not be so isolated after
`all.
`
`VI. BACK TO THE FUTURE
`
`A natural question to ask at this point is whether there
`is any remaining need to build Photown in the first place.
`If HDSL can deliver wideband service on proven copper
`transmission media, and MAN’s and DS1 circuit switch-
`ing systems can switch that service efficiently, is there
`any reason for telephone companies to work their way
`through the learning curve for optical transmission and
`more advanced switched services such as broadband
`ISDN?
`HDSL provides a good vehicle for the widespread de-
`ployment of DS1 services, but it is pressing the limits of
`transmission over twisted-pair wire. Fiber optic transmis-
`sion, once installed in an area, will always offer a cheaper
`upgrade to DS1 service, and its ability to carry signals
`hundreds of times faster than DSl makes it a better long-
`term investment. As fiber loop technology reaches cost
`parity with copper for POTS, it would be a mistake not
`to use it in new construction and major repair. As the de-
`mand for advanced DS1 services develops, the economics
`will tilt more strongly in favor of fiber, eventually justi-
`fying early replacement of copper with fiber in areas that
`would not otherwise qualify for traditional loop rehabili-
`tation.
`HDSL does not make Photown obsolete; it makes Pho-
`town practical. The development of a large-scale market
`for switched DS1 services will similarly accelerate the de-
`ployment of 150-600 Mb/s broadband ISDN switches.
`The first use of 150 Mb switching will probably be for
`tandem switching of multiple DS 1 channels concentrated
`by MAN’s. Once deployed, these switches may also pro-
`vide broadband ISDN services for leading-edge subscri-
`bers.
`The network of the 21st Century will have a number of
`the characteristics that we have seen in the popular fore-
`casts of the last several years. Broadband ISDN services,
`
`

`
`768
`
`
`
`IEEE JOURNAL ( I N SELECTED AREAS IN COMMUNICATIONS. VOL. 9. NO. 6, AUGUST 1991
`
`delivered via fiber-optic local loops, will integrate a broad
`range of voice, data, image, and video services to the
`business and (eventually) residence markets. Universal
`broadband communications will provide the basis for a
`rich market in information and entertainment services.
`Copper loops with HDSL technology are a necessary
`evolutionary step towards a broadband fiber network.
`Wideband copper transmission is not the ultimate network
`of the future. It is the bridge that will get us there.
`
`Wil Walkoe received the B A degree from Kal-
`amazoo College, Kalamazoo, MI, in 1963, and
`the M S. degree in industrial engineering and
`Ph.D degree in mathematics from the University
`of Wisconsin in 1969 and 1981, respectively
`He is a Senior Member of Technical Staff in
`Systems Engineering, Ameritech Services, Roll-
`ing Meadows, 1L. Prior to joining Ameritech in
`1988, he taught for eleven years at Grand Valley
`University, Allendale, MI, and worked in AT&T'5
`Data Systems Division for seven years in systems
`engineering and product management
`
`Thomas J. J. Starr (S'74-M'74-M'83)
`received
`the B.S. degree in computer engineering, and the
`M.S. degree in computer science from the Uni-
`versity of Illinois, Champaign.
`He was a Member of Technical Staff at AT&T
`Bell Labs in Naperville, IL for 12 years, where he
`worked on IAESS circuit design, memory updat-
`ing systems, SESS ISDN circuit design, system
`architecture, and ISDN standards (ISDN basic
`rate, layer I ) . He joined Americh Services, Inc.
`in May 1988, and is now a Senior Member of
`Technical Staff in the newly formed Advanced Network Technologies or-
`ganization. Shortly after joining ASI. he was elected Chairperson of the
`United States Exchange Carriers Standards Association Standards Working
`Group T l E l . 4 , which is responsible for layer-I network interface stan-
`dards for ISDN basic rate (the U-interface (2BlQ) and SIT interfaces).
`This group also has a project to study high-bit-rate digital subscriber line
`(HDSL) technology which will permit nonrepeatered 1 .S Mb/s service over
`existing nonconditioned CSA wire loops. He leads a project team within
`Ameritech to study this technology. He is also investigating next-genera-
`tion switching architectures, which include refinement of ISDN and plan-
`ning towards 1.5 M b / s (and N X 64 kb/s) switching. He participates as
`a U.S. delegate to the CCITT SG-XVIII International Standards body to
`help carry the U.S. standards positions into the intemational standards
`bodies. He holds five U.S. patents.