xDSL Access Networks
Residential Broadband, Second Edition
Author: George Abe
Publisher: Cisco Press (53)
This chapter covers the following topics:
Current Telco Services
Digital Loop Carrier Overview
Very High Data Rate DSL
High Data Rate DSL
ISDN Digital Subscriber Line
Single-Line Digital Subscriber Line
Early Provisioning for xDSLs
Factors in Choosing Which DSL to Offer
Technical Challenges to xDSL
Industry Challenges to xDSL
Cable operators offer their HFC networks for RBB services. How are the telephone
companies to counter? Some telcos have toyed with the idea of building HFC
networks of their own or buying cable operators. Among these have been PacBell,
Telecom Australia, and Ameritech. However, most telcos prefer to capitalize
on their existing multibillion-dollar embedded base of subscriber lines by
pushing the evolution of a new technology called digital subscriber
line (DSL). The subscriber line,
often referred to as the local loop, is a pair of
wires that connects each subscriber to a local telephone building called the central
office (CO). DSL uses existing subscriber lines to transmit data
at speeds of over 100 or even 1000 times current modem service. Hence, it
is the vehicle by which telephone companies can compete against cable for
video and high-speed data services.
Before proceeding, we note a semantic distinction between loop
and line. DSL refers to digital subscriber line. For
the purposes of this chapter and book, loop refers to
a physical wire pair between the home and the carrier premises, either a central
office or remote terminal. A line is a service concept
that denotes a capability purchased by a customer. Hence, xDSL refers to line
because it is a service concept.
DSL is deployed by U.S. telephone operators to modernize their
wiring to households. DSL service extends fiber deeper into the neighborhood
and closer to the consumer. U.S. phone companies have enormous maintenance
requirements for their 171 million wired connections to customers. Regulatory
relief and business practice make more than $20 billion per year available
for telco maintenance budgets. In the maintenance process, certain upgrades
are permissible that have the effect of increasing the bit rate and lowering
To make matters complicated, multiple flavors of DSL technologies are
emerging. The various DSL technologies discussed in this chapter are high
data rate DSL (HDSL), asymmetric
DSL (ADSL), single-line DSL
(SDSL), ISDN digital subscriber
line (IDSL), and very high data
rate DSL (VDSL). Collectively,
these are referred to as xDSL. Before considering xDSL technologies, a quick
look at the history and architecture of the current telco services will be
To date, xDSL services trail cable modem deployment. At the March 1999 meeting
of the ADSL Forum, it was reported that, as of December 1998, there were 45,000
domestic U.S. ADSL customers, compared with more than 500,000 cable modem
subscribers (330,000 with @Home alone) and
75,000 customers worldwide.
Conventional wisdom is that phone wire is too restrictive for high-speed
service because customers are accustomed to phone conversations and 56 Kbps
modem service. However, the basic characteristics of phone wire have
the capacity to carry millions of bits per second. The reason they don't is
that present voice service is provided in analog mode and filters are installed
on phone interfaces and loops that suppress signals above 3400 Hz. Given these
filters, the data-carrying capacity of phone wire is limited to 56 Kbps, using
the latest generation of V.90 modems. Other services, operating in a digital
mode, occupy bandwidth such as those shown in Table
Bandwidth-limiting techniques in the voice network enable economical frequency multiplexing over long-distance lines. If users could
have 20,000 Hz per phone line, which is the frequency response of a modern
stereo system, bandwidth on the backbone would be reduced by a factor of 6.
In addition, as noted in Chapter 2, “Technical
Foundations of Residential Broadband,” pushing very high frequency
through wires creates problems of attenuation and distortion
that negatively impact the integrity of the signal.
Today, new modulation, equalization, and error-control techniques make
subscriber lines capable of transmission rates greater
than 6 Mbps for distances up to 3 miles. How much greater depends on the length
of the phone wire between the subscriber and the telephone office, the physical
condition of the wires (corrosion, insulation, bridged taps, and crosstalk),
and the thickness of the wires (the thicker the wire, the less the resistance
and the greater the distance served). Table 4-2
shows the trade-off of distance and speed over 24-gauge phone wire (0.5 mm
Table 4-2. Distance Versus Speed over 24-Gauge Phone Wire
Source: Alcatel at frcatel.utc.sk/hwb/ta_AWG.html
Before examining how technology is pushing this envelope, it will be useful
to overview two current telephone servicesplain old telephone services
(POTS) and Integrated Services Digital
Networks (ISDN)as well as the rationale for
Telephone switching equipment, which establishes phone connections, is located in the CO. Customers
are connected to the CO over thin-wire pairs, also referred to as local loops. These thin-wire pairs
are segmented in lengths of 500 feet. Lengths are spliced together as needed
to reach from the CO to the customer's home.
The first 500-foot segment from the CO, 26-gauge wire, is normally 0.41
mm in diameter. This is the thinnest type of phone wire. After the first segment,
the phone wire is often a thicker diameter, such as 24-gauge (0.50 mm) or
the thickest, which is a 19-;;gauges
(0.82 mm) wire. Because resistance is inversely related to thickness, the
thicker-gauge wire is reserved for customers who are farther from the CO.
Resistance in the wire is a function of temperature as well as wire thickness.
At 70° Fahrenheit, 26-gauge wire has about 40 ohms of resistance per thousand
feet; 19-gauge wire induces 10 ohms per thousand feet.
Local loops are bundled in large cables called binder
groups. The job of the field technician is to cross-connect a drop
wire to your home and then to a phone wire in a binder group. Fifty phone
wires to a binder group is a typical configuration near the subscriber.
Some binder groups contain as few as 20 local loops, and others contain
as many as a few hundred. Feeder cablesroughly the first 9000 feet coming
out of a central officecan have hundreds or even thousands of pairs
bundled together. These are referred to as binder groups as well. Figure
4-1 shows a simplified connection between a CO and customers.
The copper in phone lines, although very thin, adds up when considered
network-wide. U.S. telcos are the largest single consumer of copper in the
world. The local loop consumes nearly 50 percent of telco capital cost; that
is capital cost of a local loop is $1500 to $2000 with about 50 percent being
used for materials, including copper.
Therefore, a premium is put on having the thinnest possible copper, consistent
with transmission fidelity. Thinner wires mean less real estate and fewer
material costs. Thinner wires are lighter, also reducing the cost of aerial
runs. Generally, minimizing the consumption of copper is a goal in the design
of phone systems and is one reason for telcos' interest in fiber technologies.
Eventually, a single-wire pair is
extracted from feeder cables and distribution
cables of fewer wire pairs until the single-wire pair is connected to your
house over the drop wire. The drop wire connects
to your home in a junction box called the network interface
device (NID). The NID is typically a passive device that serves
external phone wire to internal phone wire in the home.
Basic Rate ISDN (BRI) is a digital service that
Kbps over phone wire and up to 18,000 feet of 24-gauge wire. Its standard
implementation (ANSI T1.601 or ITU I.431) employs echo cancellation to separate
the transmit signal from the received signal on a single pair of wires. It
uses bandwidth from 0 to about 80 kHz.
European systems use 120 kHz of bandwidth. Therefore, provisioning of ISDN
and analog POTS on the same local loop is not possible because both services
utilize frequencies less than 3400 Hz.
This is not a big deal in the United States because there are less than
1 million ISDN lines provisioned. However, in Europe, ISDN is more widely
deployed and therefore there needs to be a coexistence and migration strategy
to move ISDN users to xDSL. In Europe, there are more than 6 million ISDN
lines, most using a form of ISDN that uses 0 to 120 kHz. The intent is
to migrate ISDN users without changing their local loops. The current
idea within the European
Telecommunications Standards Institute (ETSI) is to have ADSL
start at 140 kHz and proceed upward and to use an ISDN splitter instead
of a POTS splitter. This puts ADSL on a different frequency plan than
in the United States.
Certainly there is new interest in high-speed services
over existing subscriber lines. However, even low-speed data service using
telephone modems creates new problems for the phone system. These limitations
associated with traditional telco networks are also driving interest in xDSL
Currently, there are more than 22,000 COs in the United States, serving
171 million lines. Hundreds more offices are served by long-distance carriers
(AT&T, MCI, Sprint, and so on), also referred to as interexchange carriers (IXC).
The average CO serves 7600 lines; a few serve as many as 100,000 lines. The
problem of limited space in conduits in and out of COs is becoming more severe
as consumers add second lines for home use. About 20 percent of U.S. homes
have added second lines for use in home offices and for talkative teenagers.
In Beverly Hills 90210, there is an average of nearly 4 subscriber lines per
household (that is, main number, fax, teenager, maid).
Central offices are big, expensive buildings. Reducing real estate needs
by using more compact electronics can account for noticeable savings. In places
where real estate is at a premium, such as Rome and Tokyo, the sale of telco
property can generate enough money to fund digital buildouts of telephone
services. In other words, telcos can trade buildings and real estate for new
digital infrastructures. Therefore, an incentive exists to move to more compact
facilities. One way to facilitate such a move is by installing fiber for new
services and distributed switching systems, hence the move to DSLs.
Users obtain services by connecting to the CO. Their distance from the
CO dictates the cost of providing the service and, in some cases, the type
of service received.
Two standardized range limits exist. Revised resistance design rules (RRD)
limit distance to 18,000 feet for 24-gauge and 15,000 feet for 26-gauge wire.
RRD rules are used for ISDN. Another range limit, carrier serving area (CSA)used
for HDSL servicelimits service to 12,000 feet for 24-gauge and 9000
feet for 26-gauge wire. Half of U.S. lines are within 9000 feet of a CO or
remote terminal; 80 percent are within 15,000 feet. (The RRD and CSA distances
are estimates because the actual distance is a function of line quality. If
a particular local loop has severe impairments, the distances will be shorter
than these estimates.)
Internet service over POTS and ISDN
lines is readily offered because of the ubiquity of phone service and the
widespread availability of modems and voice switches provisioned for ISDN.
The problem for telcos is that both POTS and ISDN use the CO voice switch,
and data has a different usage profile from voice.
The major difference is that data sessions are longer than voice calls.
A 1996 study conducted by U.S. West showed that voice calls are an average
5.64 minutes in length. Calls to Internet service providers (ISPs) are an
average 32.47 minutessix times the length of a voice call. Longer sessions
might mean that the number of ports on the switches is insufficient, causing
busy signals. When busy signals become excessive, the telco is obliged to
buy more switching equipment. At $500 to $1000 per line, this is an expensive
proposition. The economic and practical impacts of these estimates are significant,
particularly when extrapolated to reflect the continuing growth of Internet
The difficulties associated with this discrepancy in call length are
particularly noticeable in areas of the country that have a lot of Internet
dial-up activity, such as California. Because of the inordinate costs to their
infrastructure caused by data sessions, Pacific Telesis (now SBC) has asked
that ISPs be subject to access fees.
Telcos are searching for ways in which data traffic can be offloaded
from voice switches to specialized data communications equipment. xDSL services
are a method to do this. This is the basic difference between ISDN and xDSL:
ISDN goes through a voice switch; xDSL bypasses it. This makes rollout costs
incremental for xDSL, whereas ISDN requires telcos to upgrade their voice
switches (at a cost of up to $500,000 a pop) before ISDN can be offered. Therefore,
xDSL can be considered a lower-cost platform for data service, even
though it offers faster bit rates than ISDN.
One of the primary architects of OpenCable, Michael
Adams, explains the key concepts of this initiative in his book
Broadband, Second Edition
by George Abe
Introduces the topics surrounding high-speed networks
to the home. It is written for anyone seeking a broad-based familiarity
with the issues of residential broadband (RBB) including product
developers, engineers, network designers, business people, professionals
in legal and regulatory positions, and industry analysts.