The Internet is a network of networks that interconnects computers around the
world, supporting both business and residential users. In 1994, a multimedia
Internet application known as the World Wide Web became popular. The higher
bandwidth needs of this application have highlighted the limited Internet access
speeds available to residential users. Even at 28.8 Kilobits per second
(Kbps)—the fastest residential access commonly available at the time of this
writing—the transfer of graphical images can be frustratingly slow. This
report examines two enhancements to existing residential communications
infrastructure: Integrated Services Digital Network (ISDN), and cable television
networks upgraded to pass bi-directional digital traffic (Cable Modems). It
analyzes the potential of each enhancement to deliver Internet access to
residential users. It validates the hypothesis that upgraded cable networks can
deliver residential Internet access more cost-effectively, while offering a
broader range of services. The research for this report consisted of case
studies of two commercial deployments of residential Internet access, each
introduced in the spring of 1994: · Continental Cablevision and Performance
Systems International (PSI) jointly developed PSICable, an Internet access
service deployed over upgraded cable plant in Cambridge, Massachusetts; ·
Internex, Inc. began selling Internet access over ISDN telephone circuits
available from Pacific Bell. Internex's customers are residences and small
businesses in the "Silicon Valley" area south of San Francisco,
California. 2.0 The Internet When a home is connected to the Internet,
residential communications infrastructure serves as the "last mile" of
the connection between the home computer and the rest of the computers on the
Internet. This section describes the Internet technology involved in that
connection. This section does not discuss other aspects of Internet technology
in detail; that is well done elsewhere. Rather, it focuses on the services that
need to be provided for home computer users to connect to the Internet. 2.1 ISDN
and upgraded cable networks will each provide different functionality (e.g. type
and speed of access) and cost profiles for Internet connections. It might seem
simple enough to figure out which option can provide the needed level of service
for the least cost, and declare that option "better." A key problem
with this approach is that it is difficult to define exactly the needed level of
service for an Internet connection. The requirements depend on the applications
being run over the connection, but these applications are constantly changing.
As a result, so are the costs of meeting the applications' requirements. Until
about twenty years ago, human conversation was by far the dominant application
running on the telephone network. The network was consequently optimized to
provide the type and quality of service needed for conversation. Telephone
traffic engineers measured aggregate statistical conversational patterns and
sized telephone networks accordingly. Telephony's well-defined and stable
service requirements are reflected in the "3-3-3" rule of thumb relied
on by traffic engineers: the average voice call lasts three minutes, the user
makes an average of three call attempts during the peak busy hour, and the call
travels over a bidirectional 3 KHz channel. In contrast, data communications are
far more difficult to characterize. Data transmissions are generated by computer
applications. Not only do existing applications change frequently (e.g. because
of software upgrades), but entirely new categories—such as Web browsers—come
into being quickly, adding different levels and patterns of load to existing
networks. Researchers can barely measure these patterns as quickly as they are
generated, let alone plan future network capacity based on them. The one
generalization that does emerge from studies of both local and wide- area data
traffic over the years is that computer traffic is bursty. It does not flow in
constant streams; rather, "the level of traffic varies widely over almost
any measurement time scale" (Fowler and Leland, 1991). Dynamic bandwidth
allocations are therefore preferred for data traffic, since static allocations
waste unused resources and limit the flexibility to absorb bursts of traffic.
This requirement addresses traffic patterns, but it says nothing about the
absolute level of load. How can we evaluate a system when we never know how much
capacity is enough? In the personal computing industry, this problem is solved
by defining "enough" to be "however much I can afford
today," and relying on continuous price-performance improvements in digital
technology to increase that level in the near future. Since both of the
infrastructure upgrade options rely heavily on digital technology, another
criteria for evaluation is the extent to which rapidly advancing technology can
be immediately reflected in improved service offerings. Cable networks satisfy
these evaluation criteria more effectively than telephone networks because: ·
Coaxial cable is a higher quality transmission medium than twisted copper wire
pairs of the same length. Therefore, fewer wires, and consequently fewer pieces
of associated equipment, need to be installed and maintained to provide the same
level of aggregate bandwidth to a neighborhood. The result should be cost
savings and easier upgrades. · Cable's shared bandwidth approach is more
flexible at allocating any particular level of bandwidth among a group of
subscribers. Since it does not need to rely as much on forecasts of which
subscribers will sign up for the service, the cable architecture can adapt more
readily to the actual demand that materializes. · Telephony's dedication of
bandwidth to individual customers limits the peak (i.e. burst) data rate that
can be provided cost-effectively. In contrast, the dynamic sharing enabled by
cable's bus architecture can, if the statistical aggregation properties of
neighborhood traffic cooperate, give a customer access to a faster peak data
rate than the expected average data rate. 2.2 Why focus on Internet access?
Internet access has several desirable properties as an application to consider
for exercising residential infrastructure. Internet technology is based on a
peer-to-peer model of communications. Internet usage encompasses a wide mix of
applications, including low- and high- bandwidth as well as asynchronous and
real-time communications. Different Internet applications may create varying
degrees of symmetrical (both to and from the home) and asymmetrical traffic
flows. Supporting all of these properties poses a challenge for existing
residential communications infrastructures. Internet access differs from the
future services modeled by other studies described below in that it is a real
application today, with growing demand. Aside from creating pragmatic interest
in the topic, this factor also makes it possible to perform case studies of real
deployments. Finally, the Internet's organization as an "Open Data
Network" (in the language of (Computer Science and Telecommunications Board
of the National Research Council, 1994)) makes it a service worthy of study from
a policy perspective. The Internet culture's expectation of interconnection and
cooperation among competing organizations may clash with the monopoly-oriented
cultures of traditional infrastructure organizations, exposing policy issues. In
addition, the Internet's status as a public data network may make Internet
access a service worth encouraging for the public good. Therefore, analysis of
costs to provide this service may provide useful input to future policy debates.
3.0 Technologies This chapter reviews the present state and technical evolution
of residential cable network infrastructure. It then discusses a topic not
covered much in the literature, namely, how this infrastructure can be used to
provide Internet access. It concludes with a qualitative evaluation of the
advantages and disadvantages of cable-based Internet access. While ISDN is
extensively described in the literature, its use as an Internet access medium is
less well-documented. This chapter briefly reviews local telephone network
technology, including ISDN and future evolutionary technologies. It concludes
with a qualitative evaluation of the advantages and disadvantages of ISDN-based
Internet access. 3.1 Cable Technology Residential cable TV networks follow the
tree and branch architecture. In each community, a head end is installed to
receive satellite and traditional over-the-air broadcast television signals.
These signals are then carried to subscriber's homes over coaxial cable that
runs from the head end throughout the community Figure 3.1: Coaxial cable
tree-and-branch topology To achieve geographical coverage of the community, the
cables emanating from the head end are split (or "branched") into
multiple cables. When the cable is physically split, a portion of the signal
power is split off to send down the branch. The signal content, however, is not
split: the same set of TV channels reach every subscriber in the community. The
network thus follows a logical bus architecture. With this architecture, all
channels reach every subscriber all the time, whether or not the subscriber's TV
is on. Just as an ordinary television includes a tuner to select the
over-the-air channel the viewer wishes to watch, the subscriber's cable
equipment includes a tuner to select among all the channels received over the
cable. 3.1.1. Technological evolution The development of fiber-optic
transmission technology has led cable network developers to shift from the
purely coaxial tree-and-branch architecture to an approach referred to as Hybrid
Fiber and Coax(HFC) networks. Transmission over fiber-optic cable has two main
advantages over coaxial cable: · A wider range of frequencies can be sent over
the fiber, increasing the bandwidth available for transmission; · Signals can
be transmitted greater distances without amplification. The main disadvantage of
fiber is that the optical components required to send and receive data over it
are expensive. Because lasers are still too expensive to deploy to each
subscriber, network developers have adopted an intermediate Fiber to the
Neighborhood (FTTN)approach. Figure 3.3: Fiber to the Neighborhood (FTTN)
architecture Various locations along the existing cable are selected as sites
for neighborhood nodes. One or more fiber-optic cables are then run from the
head end to each neighborhood node. At the head end, the signal is converted
from electrical to optical form and transmitted via laser over the fiber. At the
neighborhood node, the signal is received via laser, converted back from optical
to electronic form, and transmitted to the subscriber over the neighborhood's
coaxial tree and branch network. FTTN has proved to be an appealing architecture
for telephone companies as well as cable operators. Not only Continental
Cablevision and Time Warner, but also Pacific Bell and Southern New England
Telephone have announced plans to build FTTN networks. Fiber to the neighborhood
is one stage in a longer-range evolution of the cable plant. These longer-term
changes are not necessary to provide Internet service today, but they might
affect aspects of how Internet service is provided in the future. 3.2 ISDN
Technology Unlike cable TV networks, which were built to provide only local
redistribution of television programming, telephone networks provide switched,
global connectivity: any telephone subscriber can call any other telephone
subscriber anywhere else in the world. A call placed from a home travels first
to the closest telephone company Central Office (CO) switch. The CO switch
routes the call to the destination subscriber, who may be served by the same CO
switch, another CO switch in the same local area, or a CO switch reached through
a long- distance network. Figure 4.1: The telephone network The portion of the
telephone network that connects the subscriber to the closest CO switch is
referred to as the local loop. Since all calls enter and exit the network via
the local loop, the nature of the local connection directly affects the type of
service a user gets from the global telephone network. With a separate pair of
wires to serve each subscriber, the local telephone network follows a logical
star architecture. Since a Central Office typically serves thousands of
subscribers, it would be unwieldy to string wires individually to each home.
Instead, the wire pairs are aggregated into groups, the largest of which are
feeder cables. At intervals along the feeder portion of the loop, junction boxes
are placed. In a junction box, wire pairs from feeder cables are spliced to wire
pairs in distribution cables that run into neighborhoods. At each subscriber
location, a drop wire pair (or pairs, if the subscriber has more than one line)
is spliced into the distribution cable. Since distribution cables are either
buried or aerial, they are disruptive and expensive to change. Consequently, a
distribution cable usually contains as many wire pairs as a neighborhood might
ever need, in advance of actual demand. Implementation of ISDN is hampered by
the irregularity of the local loop plant. Referring back to Figure 4.3, it is
apparent that loops are of different lengths, depending on the subscriber's
distance from the Central Office. ISDN cannot be provided over loops with
loading coils or loops longer than 18,000 feet (5.5 km). 4.0 Internet Access
This section will outline the contrasts of access via the cable plant with
respect to access via the local telephon network. 4.1 Internet Access Via Cable
The key question in providing residential Internet access is what kind of
network technology to use to connect the customer to the Internet For
residential Internet delivered over the cable plant, the answer is broadband LAN
technology. This technology allows transmission of digital data over one or more
of the 6 MHz channels of a CATV cable. Since video and audio signals can also be
transmitted over other channels of the same cable, broadband LAN technology can
co-exist with currently existing services. Bandwidth The speed of a cable LAN is
described by the bit rate of the modems used to send data over it. As this
technology improves, cable LAN speeds may change, but at the time of this
writing, cable modems range in speed from 500 Kbps to 10 Mbps, or roughly 17 to
340 times the bit rate of the familiar 28.8 Kbps telephone modem. This speed
represents the peak rate at which a subscriber can send and receive data, during
the periods of time when the medium is allocated to that subscriber. It does not
imply that every subscriber can transfer data at that rate simultaneously. The
effective average bandwidth seen by each subscriber depends on how busy the LAN
is. Therefore, a cable LAN will appear to provide a variable bandwidth
connection to the Internet Full-time connections Cable LAN bandwidth is
allocated dynamically to a subscriber only when he has traffic to send. When he
is not transferring traffic, he does not consume transmission resources.
Consequently, he can always be connected to the Internet Point of Presence
without requiring an expensive dedication of transmission resources. 4.2
Internet Access Via Telephone Company In contrast to the shared-bus architecture
of a cable LAN, the telephone network requires the residential Internet provider
to maintain multiple connection ports in order to serve multiple customers
simultaneously. Thus, the residential Internet provider faces problems of
multiplexing and concentration of individual subscriber lines very similar to
those faced in telephone Central Offices. The point-to-point telephone network
gives the residential Internet provider an architecture to work with that is
fundamentally different from the cable plant. Instead of multiplexing the use of
LAN transmission bandwidth as it is needed, subscribers multiplex the use of
dedicated connections to the Internet provider over much longer time intervals.
As with ordinary phone calls, subscribers are allocated fixed amounts of
bandwidth for the duration of the connection. Each subscriber that succeeds in
becoming active (i.e. getting connected to the residential Internet provider
instead of getting a busy signal) is guaranteed a particular level of bandwidth
until hanging up the call. Bandwidth Although the predictability of this
connection-oriented approach is appealing, its major disadvantage is the limited
level of bandwidth that can be economically dedicated to each customer. At most,
an ISDN line can deliver 144 Kbps to a subscriber, roughly four times the
bandwidth available with POTS. This rate is both the average and the peak data
rate. A subscriber needing to burst data quickly, for example to transfer a
large file or engage in a video conference, may prefer a shared-bandwidth
architecture, such as a cable LAN, that allows a higher peak data rate for each
individual subscriber. A subscriber who needs a full-time connection requires a
dedicated port on a terminal server. This is an expensive waste of resources
when the subscriber is connected but not transferring data. 5.0 Cost Cable-based
Internet access can provide the same average bandwidth and higher peak bandwidth
more economically than ISDN. For example, 500 Kbps Internet access over cable
can provide the same average bandwidth and four times the peak bandwidth of ISDN
access for less than half the cost per subscriber. In the technology reference
model of the case study, the 4 Mbps cable service is targeted at organizations.
According to recent benchmarks, the 4 Mbps cable service can provide the same
average bandwidth and thirty-two times the peak bandwidth of ISDN for only 20%
more cost per subscriber. When this reference model is altered to target 4 Mbps
service to individuals instead of organizations, 4 Mbps cable access costs 40%
less per subscriber than ISDN. The economy of the cable-based approach is most
evident when comparing the per-subscriber cost per bit of peak bandwidth: $0.30
for Individual 4 Mbps, $0.60 for Organizational 4 Mbps, and $2 for the 500 Kbps
cable services—versus close to $16 for ISDN. However, the potential
penetration of cable- based access is constrained in many cases (especially for
the 500 Kbps service) by limited upstream channel bandwidth. While the
penetration limits are quite sensitive to several of the input parameter
assumptions, the cost per subscriber is surprisingly less so. Because the models
break down the costs of each approach into their separate components, they also
provide insight into the match between what follows naturally from the
technology and how existing business entities are organized. For example, the
models show that subscriber equipment is the most significant component of
average cost. When subscribers are willing to pay for their own equipment, the
access provider's capital costs are low. This business model has been
successfully adopted by Internex, but it is foreign to the cable industry. As
the concluding chapter discusses, the resulting closed market structure for
cable subscriber equipment has not been as effective as the open market for ISDN
equipment at fostering the development of needed technology. In addition,
commercial development of both cable and ISDN Internet access has been hindered
by monopoly control of the needed infrastructure—whether manifest as high ISDN
tariffs or simple lack of interest from cable operators.