| ||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||
| Copyright © 2009. National Academy of Sciences. All rights reserved. Terms of Use and Privacy Statement |
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 151
4
Collisions Between Existing Industries
and Emerging Interned Industries
Telephong As a Case Studs
INTRODUCTION
The general-purpose nature of the Internet and its basic technology
makes it possible to provide services at least comparable to those pro-
vided by other communication systems. Perhaps the clearest such trend
today is the growing use of Internet technology by providers of telephone
service and the emergence of a variety of voice communications services
over the Internet. For reasons that range from reducing the costs of con-
ventional telephony by using less expensive technology, to bypassing the
tariff structure of the existing public switched telephone network (PSTN),
to innovating new forms of telephony services and applications, many
efforts are under way to use the Internet and its technology components
to provide voice services. These services do not merely duplicate those
provided by the PSTN; the ease with which new applications can be
introduced over the Internet opens up the possibility of a wide range of
applications involving voice and the introduction of many different en-
hancements to what we think of as telephony today.
Telephony over the public Internet is in its infancy, but telephony
over IP networks that provide appropriate provisioning or quality-of-
service technology is already viable today. Where Internet voice is de-
ployed on existing data backbones lacking supporting quality of service
and/or provisioning, such service today is generally a second-best alter-
native. However, in other cases, where IP telephony is used either with
separately provisioned bandwidth or with supporting quality-of-service
151
OCR for page 152
52
THE INTERNET'S COMING OF AGE
technologies, it has proven to be competitive with circuit-switched tech-
nologies. There are numerous small-scale examples of such deployments
for long-haul telephony as well as several instances of large-scale sys-
tems, including two Chinese telephone networks (China Telecom and
Uninet).~ Telephony over IP networks is also being deployed within
enterprises (generally replacing PBX functionality), with the goal of
achieving lower costs by operating a single network carrying both data
and voice.2
These developments are emblematic of a broader trend toward
Internet services that overlap and potentially exceed or even supersede
communications services that have long existed as distinct industries with
distinct policies and regulations governing them. Another emerging ser-
vice is Internet distribution models for audio and video, which overlap
the capabilities of traditional radio and television broadcasting and music
and video publishing and distribution. In both cases there are conflicts
between the new technologies, systems, and players made possible by the
Internet and the policies, regulations, and practices that have shaped the
extant industry. This chapter focuses on telephony, where the conflicts
have now become quite visible, but many of the issues explored here are
also relevant to these other sectors.3
WHAT IS IP TELEPHONY?
The terms "IP telephony" and "voice over IP" are used synonymously
in this report to describe the broad range of options for using Internet-
Protocol-based, packet-switched networks in support of telecommunica-
tions services of various sorts, typically for voice communications. (The
committee is careful in its use of the term "Internet telephony," because
the term implies that the public Internet is used to carry telephone calls,
which not all IP-based telephony does.) IP may be used for some or all of
the transmission and interconnectivity path as well as to provide switch-
ing, control, and services facilitation (e.g., call setup). IP telephony
encompasses services that provide connectivity among normal telephone
See Leslie Chang. 1999. "Internet Phone Service Catches on with Millions in China-
Rivals Challenge China Telecom with Cheap Rates." Wall Street Journal. December 21,
p. A14.
2The committee does not provide data on the extent of IP telephony deployment because
assembling such information is complicated by the diverse offerings, definitional questions,
and the highly dynamic nature of the business.
3This chapter is not intended as a thorough examination of technical issues underlying
Internet telephony. Discussion of some key underlying technical dimensions, such as net-
work robustness, scaling, and quality of service, is located elsewhere in this report.
OCR for page 153
TELEPHONY AS A CASE STUDY
153
sets, computers, or other computer appliances, and combinations of these
types of devices. IP telephony may be used over a privately owned net-
work, a network owned by the telephony service provider or some other
third party, or over the public Internet.
To carry voice over IP, the analog signals are digitized, (usually) com-
pressed to reduce bandwidth requirements, and then broken up into frag-
ments or packets with accompanying routing information for transport
over a packet network such as the Internet. They are transported across
one or more IP networks, making use of algorithms running on the de-
vices at each end of the call to deal with late or lost packets and possibly
also making use of quality-of-service mechanisms in the networks the call
traverses. Call quality is determined by the algorithm used to digitize
and compress the signals, the capacity of the network, and the effective-
ness of any explicit quality-of-service mechanisms that are employed.
IP telephony comes in many flavors; these may or may not involve
the public Internet and may or may not involve the PSTN. Some IP
telephony applications make use of only an IP network, whereas others
make use of both IP networks and segments of the PSTN. If the PSTN is
involved in any segment of the call, a gateway is required to translate the
data and signaling (e.g., call setup or termination instructions) associated
with the call. Another function that is generally provided in any tele-
phony application is a directory lookup service that associates a phone
number or other identifier with a particular telephone line or IP device.4
Some designs amount to little more than a substitution of IP technol-
ogy for transport within the network. For instance, IP may be substituted
for other means of transporting communications within the networks of
telephony providers; this internal substitution of IP technology for trans-
port within the network raises few policy issues that are distinguishable
from those raised by the introduction of other new technologies or the
entry of new players into the "traditional" PSTN business. Another fla-
vor is the use of IP within the private network of an enterprise, where it
replaces the elements of a traditional PBX system or other corporate voice
network. Other designs make sole use of IP for transport and Internet
application servers for control and directory services, while still others
employ hybrids that combine elements of both the Internet and the PSTN.
The set of possible sorts of services that might be labeled "IP telephony" is
highly diverse and encompasses the following:
· Network operators and transport media. For example, backbone trans-
4The PSTN provides several such mappings, including between subscriber name and
phone number What the telephone directory contains' and between phone number and
telephone line.
OCR for page 154
54
THE INTERNET'S COMING OF AGE
port within a carrier network, telephony over the public Internet, enter-
prise networks, virtual private networks, PSTN, and combinations of
these;
· End-user equipment. For example, conventional telephones, IP-
based special-purpose devices (appliances), and general-purpose com-
puters;
· Local access technology. For example, PSTN analog lines, IP data
networks, and IP over DSL;
· Interfaces and gateways between IP and PSTN elements. For example,
analog phone lines, ISDN primary rate interfaces, Switching System 7
(SS7) or none in the case of pure IP telephony; and
· Architectures. For example, telephony services may be provided by
a single vendor in a centrally managed way (perhaps over a dedicated
network) or provided as a distributed service with individual users plac-
ing calls end-to-end over the Internet, perhaps making use of services
such as directory lookup provided by a third party.
NEW AND EVOLVING ARCHITECTURES FOR TELEPHONY
To understand the technical, operational, economic, and policy issues
related to IP telephony, it is useful to understand the role of architectures
and examine some specific examples. The concept of an architecture has
been a cornerstone in the development of telecommunications systems.
An architecture first requires that the underlying system be treated in
terms of a set of commonly understood elements and that these elements
have a clearly demarcated set of functions and interfaces that allow for
combining the elements.5 An architecture is driven by two factors in
addition to the identification and selection of elements and interfaces:
technology and world view. The state of the technology places bounds on
what is achievable. The world view is the way an individual, entity, or
organization views the world, its relationship to that world, and how that
world and that relationship will evolve over time. The PSTN, for ex-
ample, with its more centralized architecture, reflects a world view differ-
ent from that of the Internet, with its more distributed design. The limits
imposed by technology are typically less confining than the limits that are
self-imposed by the designers or architects; in practice, therefore, world
view is often the more powerful driver of architecture.
5Terrence McGarty. 1sso. ''Alternative Networking Architectures: Pricing, Policy, and
competition, Information Infrastructures for the 1ssos.~, John F. Kennedy School of Gov-
ernment, Harvard university, November.
OCR for page 155
TELEPHONY AS A CASE STUDY
155
IP Telephony Architectures
A multitude of network architectures and implementations are en-
compassed by the label "IP telephony." One way to portray this diversity
is to portray it in terms of architectural classes. A four-class taxonomy,
with illustrations of each class, is presented in Figure 4.1.
Even a single class of architecture has many possible configurations.
As shown in Figure 4.2, the telco-to-telco (class 1) architecture encom-
passes two quite different ways of employing voice over IP, even though
each way makes use of a basic PSTN-IP gateway to convert telephony
signals into IP packets. The first (Figure 4.2a) consists of a dedicated
network or channel configuration that uses a private network. The opera-
tion of this design can be explained as follows. A telephone user in, for
example, New York desires to call a telephone user in Moscow. The local
telephone user in both cases will access the system by means of a standard
telephone. The user places the call to a local exchange carrier, which then
sends the call to another switch. This switch places the call through an IP
gateway node (ION) that connects locally to a router and the call proceeds
over a private network. The process is reversed on the terminating side.
The gateway provides translation from the telephony world to the IP
domain by performing three conversions between telephone signaling
and IP signaling, between the conventional voice signal and data packets,
and between telepone numbers and IP addresses (if required). The sec-
ond class 1 approach (Figure 4.2b) is the public Internet approach adopted
by a number of IP telephony start-up companies. It uses the same generic
form of entry a gateway between the PSTN and an IP data network-
but replaces packet transport over dedicated IP links with transport over
the public Internet. It reflects both a different architecture and a different
world view of IP telephony. For example, in contrast to the architecture
using a private IP network, there is no PSTN switch or function (such as a
billing system) provided by a switch.
Increasingly, one sees new applications based on a comingling of the
Internet and PSTN telephony architectures. These emerging offerings
include "click-to-speak" voice conversations launched from Web pages
and unified messaging services (which combine voice mail, e-mail, and
fax). Another application, Internet call waiting, is aimed at customers
who use a single phone line for both normal phone service and their dial-
up connection to the Internet. While the service, which uses the Internet
to inform a customer when a call is being made to the customer's phone
line, appears on the surface to be quite similar to conventional call wait-
ing, it has very different requirements behind the scenes. Implementa-
tion of conventional call waiting is internal to the telephone switches in
OCR for page 156
156
I Telco | 4= / Telco \ ~
| Device ~,~ J \ Transport / ~J
~L_.~ _
| Device | ~J \ Transport / ~J
Class 1: Telco to telco. The end users employ existing telephony equipment
through a conventional PSTN connection while IP (either private network or
Internet) is used for a portion of the connection through interconnects with the
PSTN (e.g., IP used over long-distance segments).
THE INTERNET'S COMING OF AGE
I P-Telco
Gateway
IP-Telco
Gateway
nit
Lit
Telco
Device
<~ g/ Telco \ ~Te oo Edg4
~ If/ ~
I P-Telco
Gateway
I IP Device ( ~
Class 2: Telco to IP. Conventional telephone devices are interconnected with IP
telephony devices (a computer or other IP-enabled device).
FIGURE 4.1 Four classes of IP telephony architectures. SOURCE: Taxonomy
adapted (with extensions) from David D. Clark. 1997. A Taxonomy of Internet
Telephony Applications. Proceedings of the Telecommunications Policy Research
Conference (TPRC).
OCR for page 157
TELEPHONY AS A CASE STUDY
157
IP Device ~J
IP
\Trans~
IP Device (it
Class 3: IP to IP. Computer telephony devices are used at both ends and the
connection is made solely through IP networks or the Internet. While gateways
to the PSTN would be added to provide interconnection with PSTN customers,
this architecture does not make use of any PSTN elements and thus depends on
the development of a full range of signaling services using IP only.
IP Device ~ ~
Rae; ~
IP
T
I P-Telco
Gateway
~ ~ P Edge
ransport / 4=
I P-Telco |
Gateway
Telco
\ Transport /
Class 4: IP to telco to IP. Other combinations of PSTN and public and private IP
network links. The figure illustrates the use of the PSTN to provide a connection
between two Internet "subclouds." This architecture might be used to enable IP
telephony across Internet providers in the absence of adequate quality of service
or other capabilities over an IP connection.
OCR for page 158
158
New York
Moscow
PTT
Switch
Switch
(a) Using a dedicated private network
New York
(b) Using the public Internet
THE INTERNET'S COMING OF AGE
Billing and Customer Care
System
En '
I
Router
Router
Ed
~3
~3
EM
Router
Router
I nternet
Seoul -
Company
Private
IP Network
NOC
En
ETCH
PTT
FIGURE 4.2 Two examples of class 1 IP telephony configurations. LEC, local
exchange carrier; IGN, IP gateway node; and PTT, Post, Telegraphs, and Tele-
phones.
OCR for page 159
TELEPHONY AS A CASE STUDY
159
the PSTN; in a hybrid PSTN-Internet application, switches have to com-
municate with IP routers.
Although the commercial viability of various forms of IP telephony
remains untested, the prospect of future developments makes it possible
to think in terms of a revolutionary as opposed to an evolutionary use
of IP technology, in which phone calls become just one application on the
public Internet. Not everyone agrees on how compelling the technical
and business cases are for offering voice services over the public Internet
or what the time frame for such a transition might be, but there is little
doubt that both new and existing players will enter this market. The
timing and viability of such developments depend in part on being able to
provide the transmission resources needed to obtain sufficient voice qual-
ity (either by deploying QOS mechanisms across ISP boundaries or by
taking advantage of a rising tide of capacity that makes the average qual-
ity sufficient to meet voice requirements; see Chapter 2~. The motivation
for such a shift would not be just the cost reductions associated with using
IP-based components and systems but would also be the ability to offer
new features, particularly at the human interface, made possible by a shift
to more intelligence in end-user devices.
The Evolving Architecture of the PSTN
Over the past several decades, the PSTN has also been changing, from
a network with only basic circuit-switching functionality to a circuit-
switched network architecture known as the advanced intelligent net-
work (AIN). This architecture makes use of centralized, closed-applica-
tion databases and a PSTN-specific software development environment
(known in PSTN lingo as the service creation environment) to enable a
range of capabilities such as toll-free number dialing (including routing a
call to the appropriate point), credit card calling, and flexible call for-
warding.
As the architecture evolves, the intelligence of the PSTN is becoming
increasingly distributed and open. The PSTN's architecture is evolving
from the AIN model, which uses centralized switching system databases
and interoffice signaling, into a model that uses direct signaling to more
distributed databases. There have also been efforts to further separate
the applications from the operating system (i.e., the basic switching soft-
ware) and the underlying infrastructure so as to make the PSTN a more
. .
Open service creation environment.
The opening up of the PSTN service creation architecture is aimed at
increasing the flexibility of the infrastructure and enabling a greater num-
ber of parties to create useful services. In the AIN, the service creation
environment and the applications are provided by the service provider
OCR for page 160
160
THE INTERNET'S COMING OF AGE
and generally closed to third parties. Open AIN was proposed by PSTN
operators in the mid-199Os as a way of enabling a set of entities beyond
the service provider to create services. Today, by virtue of the PSTN's
connectivity to the Internet, service creation is being enabled at the edges
of the network. These moves toward a more open environment increase
flexibility but pose challenges related to maintaining the stable operation
of the network; some of these are discussed below.
While the intelligent network is becoming more distributed and open,
there is another fundamental change under way in the PSTN: it is becom-
ing more datacentric, with high-speed data access, transport, and switch-
ing an architecture that many believe will ultimately replace the present
intelligent, circuit-switched architecture. At the same time, the transport
backbone is evolving, with greater deployment of optical transport, such
as wave division multiplexing (WDM), to accommodate the voracious
demand for bandwidth. IP over ATM over WDM or IP directly over
WDM are emerging to replace other optical communications technolo-
gies, such as SONET or SDH, as capabilities are introduced into the opti-
cal layer. Also, local access is evolving to support higher data speeds as
datacentric communications are introduced in the local loop in the form
of DSL and digital optical fiber deployments closer to end users in tele-
phone networks and in the form of hybrid fiber coax deployments in
cable networks.
PSTN's evolution from a predominantly circuit-switched to a packet-
switched architecture is producing a significant change in architecture
construct and concomitant operations. New packet-based data (ATM
and IP) capabilities are being introduced in combination with the existing
SS7.6 This transformation has included the adoption of new, packet net-
work telecommunications protocols. Along with new protocols to estab-
lish connections between IP and PSTN systems, these packet-based capa-
bilities allow voice telephony to continue to be provided even as the
network architecture moves away from the circuit-switched model. The
new architecture, protocols, and intelligence capabilities together enable
a richer set of prospective applications and services, such as Internet call-
waiting. While there is surely room for the market to experiment with
various price/quality trade-offs, customers will expect the emerging data-
centric architecture and technologies to provide the quality they associate
with today's PSTN a challenge being addressed at this time.
6In the long term, a possible direction is that the introduction of MPLS with core ATM
switching could provide the necessary infrastructure and control capabilities.
OCR for page 161
TELEPHONY AS A CASE STUDY
Architectural Contrasts Between IP Telephony and Today's PSTN
161
Although the PSTN architecture is changing in significant ways, there
remain considerable differences of approach between it and the various
architectural alternatives emerging for IP and Internet telephony. These
differences, which stem in large part from the fundamental contrast be-
tween the PSTN's centralized design and the Internet's open, distributed
architecture, include the following:
· Unlike in the PSTN, in IP telephony the function of transporting the
packets carrying voice can be carried out by an entity different from the one
providing application services (e.g., call agents and directory services). This
means that voice telephony service can exist as an overlay to IP net-
works indeed, voice traffic can be carried over networks unbeknownst
to the IP network operators. Application support servers are not even
constrained to be located within the same network as the customer or
even the same country. It also means that Internet telephony quality
depends not only on the quality of the functions (e.g., a call agent) offered
by the telephony provider itself but also on the quality (including reliabil-
ity and freedom from congestion) of all of the underlying networks over
which communications pass.
· Like other packet data, traffic associated with a phone call over the Internet
will transit one, two, or many providers depending on which networks the calling
parties are attached to and how the networks are interconnected. The route
taken by a packet is a function of numerous routing decisions made within
the network; it may change dynamically, and it is outside the control of
the end user.
· IP telephony application servers do not necessarily provide the samefunc-
tionality as the PSTN. For instance, an application server established by a
voice telephony overlay service might be concerned only with providing
directory and call setup services. Once a call is established, data packets
will only flow between the callers, and the voice telephony provider's
server will have no access to the content of voice calls. It would, for
example, be impossible in this architecture for the telephony provider
itself to carry out a wiretap order to provide the content of a call to law
enforcement. Nor would the provider itself be able to offer priority rout-
ing for emergency calls, as might be desired to provide functionality
analogous to 911 telephone calls.
· The general-purpose nature of the Internet means that it can offer voice
communication in novel ways that do not parallel classic telephony. Telephony
is understood to refer to communication between devices attached to the
public switched telephone network (e.g., telephony regulations do not
apply to two-way radios or in-building intercom systems). The line be-
OCR for page 166
OCR for page 167
OCR for page 168
OCR for page 169
OCR for page 170
OCR for page 171
OCR for page 172
OCR for page 173
OCR for page 174
OCR for page 175
OCR for page 176
Representative terms from entire chapter:
local exchange
166
THE INTERNET'S COMING OF AGE
bilities;7 the IETF Media Gateway Control working group, which is devel-
oping an architecture for controlling gateways between the IP networks;8
and the Telephone Number Mapping group, which is working on proto-
cols for mapping telephone numbers to other attributes (e.g., URLs) that
can be used to contact a resource associated with the numbers.9
· International Telecommunication Union Standardization Sector (ITU-
T). ITU-T activities include ITU-T Study Group 13,1° which is studying
IP-based network issues such as interoperability with other networks,
signaling requirements, numbering, and security, and Study Group 16,
which is working on protocols and standards for multimedia services.
· European Telecommunications Standards Institute. ETSI, the devel-
oper of the GSM standard for cellular telephones, has a project, Telecom-
munications and Internet Protocol Harmonization Over Networks
(TIPHON), that is developing an architecture and requirements for inter-
operability and exploring technical aspects of billing; call control; nam-
ing, numbering, and addressing; and quality of service.ll
· Softswitch Consortium. This industry group provides interoperabil-
ity testing facilities, testing events, specifications, reference implementa-
tions, and development resources for a number of voice and multimedia
communications standards.l2
As this list suggests, there are areas of overlap and conflict among the
activities of these groups. These have been diminishing, however, and
there are instances of collaboration between groups. For example, the
IETF Media Gateway Control working group is collaborating with ITU-T
Study Group 13.
Central to a successful convolution is the resolution of interoperability
issues in numbering and addressing; signaling and control and service
creation capabilities; and robustness concerns. Each area is treated in turn
below.
7See IETF IP Telephony working group charter, available online at
TELEPHONY AS A CASE STUDY
167
Addressing and Number Portability
No matter what the ultimate end point, telephone number assign-
ment will be more complicated as IP telephony services emerge. Number
portability across the public switched telephone networkl3 and IP tele-
phony services is an important consideration, at least over the short term.
Today, local phone number portability in the PSTN is provided by the
local exchange carriers, permitting customers to switch local carriers with-
out changing their standard PSTN phone number.l4 In the future, there
may be no conventional telephone number associated with customers
using IP telephony; in fact, depending on how their IP address is as-
signed, they may or may not have a fixed IP address. How will number
portability be handled? Several requirements for a smooth evolution
stand out. First, customers should be able to in some fashion transfer
("port") their existing telephone number to the Internet-based service,
just as customers can today retain, at the same location, the same tele-
phone number when they switch from one local exchange carrier to an-
other, allowing them to continue to receive calls directed to the phone
number provided by their original local exchange carrier. Second, there
is the question of how someone on the PSTN calls an IP telephony sub-
scriber. This suggests that Internet service providers and other Internet-
based telephony providers should be able to issue new telephone num-
bers to their customers, even to those who do not have conventional
telephone service, so as to provide compatibility across calls originated or
terminated on conventional and IP-based telephones.
Additional issues arise when a given household uses both PSTN and
Internet services. For example, what address or number should be used
when a standard PSTN (E.164) number is assigned to a household that
uses an IP address (perhaps from an IP telephony directory server that
maps a name or other identifier to an address) for an IP telephony appli-
ance? Portability needs to address both aspects for subscribers who wish
to port their service as well as for subscribers using IP who wish to change
their Internet service providers.
13E.164 is the ITU standard ("Recommendation") for the international public telecommu-
nications numbering plan, which specifies a geographically hierarchical numbering plan in
which numbers are assigned to customers by carriers.
14Local number portability was mandated by the Telecommunications Act of 1996. In its
rules implementing the act, the FCC allowed local telephone companies to assess charges to
recover the costs of implementing and providing portability both a charge to be paid by
other carriers in exchange for the use of number portability facilities and a monthly charge
for all telephone customers.
168
THE INTERNET'S COMING OF AGE
Beyond these more immediate interoperability considerations lie
broader questions of addressing. For example, telephone services today
already allow using abstractions of the phone numbers. Voice-activated
dialing allows a caller to substitute a phrase (e.g., "call Mom") for a phone
number. Other services offer a single number that then follows the cus-
tomer to whatever phone is in use at the time a call is put through. Such
abstractions somewhat like the level of indirection provided by today's
Domain Name System, which allows access to Internet devices via a name
rather than a numerical address could also be applied to the IP tele-
phony domain.~5 In the long term, phone numbers could be replaced by
other identifiers. How to provide suitable directory services is currently
being explored in the forums listed at the start of this section; agreement
on a standard will be crucial to building hybrid Internet-PSTN or Internet-
only telephony networks that appear seamless to the end user.
While this report does not explore these issues in depth, the issues
that arise when one starts to explore solutions for all of these name/
number portability requirements closely resemble those that arise for the
Domain Name System (see "Scaling of the Internet's Naming Systems" in
Chapter 2~. For example, there are similar issues of ensuring scalability,
supporting dynamic updating, ensuring that the name/number infra-
structure is robust and secure, and authenticating updates to directory
entries.
Signaling and Control and Service Creation
As architectures for IP telephony emerge, there are unresolved ques-
tions of how they will interface with the existing SS7-based global PSTN,
including how signaling and control, the functions that allow calls to be
set up and the network to be managed, will be provided for a hybrid
PSTN/Internet infrastructure. Solutions must be sufficiently standard-
ized to allow interoperation and yet flexible enough to encourage innova-
tion in new applications. Open questions include the following: What
signaling capabilities need to be established to provide messaging across
all media (wireless and wireline) and different types of devices? What is
required to meet the needs of telephony that runs over the public Internet
in contrast with telephony that runs over private networks using IP tech-
nology, and what are the implications for the signaling and control infra-
~5Unlike the old naming schemes for telephone exchanges te.g., '~Jacksonville 6-500 i, in
which there was a fixed mapping between name and exchange, a naming scheme that
provides indirection allows the number with which a name is associated to be easily
changed, perhaps at the direction of the telephone user.
TELEPHONY AS A CASE STUDY
169
structure to support these changes? How is quality of service assured
across the multiple providers for data services? These are some of the
important questions that the working groups described above are ad-
dressing.
The control capabilities must be able to handle both the pure IP tele-
phony and the interdomain calls between Internet telephony and PSTN
elements. The committee foresees some interesting complications. For
instance, a household of the future might well employ both conventional
PSTN telephone and IP telephony equipment, with the latter making use
of one or more local access technologies (e.g., IP over DSL or IP over a
cable modem). Telephones of the future might use a variety of local
access technologies, such as IP over ATM or over DSL. It will need to be
possible to manage each of these capabilities with the appropriate proto-
cols. Another example of these complications is the click-to-speak service
mentioned earlier, which may involve a voice connection between the
customer and the service representative or the downloading of streaming
video from a server to the prospective customer. A suite of open standard
protocols will be needed to enable these interoperable sets of PSTN/
Internet services.
Robustness
The robustness of the PSTN benefited in the past from agreement
among a relatively small number of players to follow a tightly defined
architecture and set of operating practices. As the PSTN market has
grown, the number of players has grown, too, potentially affecting ro-
bustness. However, the opening up of the PSTN architecture and inter-
operation with other IP networks give rise to robustness concerns that go
beyond those that would be posed by conventional new entrants to the
PSTN. These concerns fall into two categories. First, there are concerns
about how to ensure the reliability of a more open architecture in which
there are fewer controls over the inputs, e.g., signaling and control mes-
sages, received by PSTN networks and in which these inputs may inter-
act. Second, there are concerns about how to ensure overload controls
and network availability in this more open environment.
As a network is opened up, additional attention must be paid to
authentication (Is the originator of the message authorized to perform the
function?) and validation (Is a request a reasonable one that can be ex-
ecuted without harm to the network of signaling and control messages?.
Another issue is how to localize the impact of problems. How can a more
open service creation environment be designed to create new applications
without adversely affecting the applications of others? What mediation
capabilities and other new service-creation or application programming
170
THE INTERNET'S COMING OF AGE
interfaces should be developed? Can there be service creation at the
edges without having to worry about multiple-creator application inter-
actions and adverse consequences? If so, how does this manifest itself in
an Internet/PSTN environment a hybrid of centralized or partially dis-
tributed service creation in the PSTN plus the fully distributed service
creation of the Internet? How can the combination of SCEs flourish in this
evolution? How do the provider-proprietary applications for specific
customers affect the service creation environments?
If networks carrying voice networks are to continue to meet stringent
performance requirements in the face of opening them to a diverse set of
providers, the ability to control against overload and provide enhanced
robustness will need to be included. These mechanisms will need to be
designed not only into the systems comprising the PSTN/Internet combi-
nation but also into the basic signaling and control architecture, to pre-
vent serious congestion, service degradation, and severe outages.
IMPLICATIONS OF IP TELEPHONY
FOR TELEPHONY REGULATION
The emergence of IP telephony heralds conflicts between, on the one
hand, IP telephony's practices and assumptions and, on the other hand,
the practices and assumptions of the existing regulatory regime. These
conflicts stem in large part from the contrast between the dynamic, rapid
change that Internet-based innovation enables and the historically rela-
tively stable nature of PSTN technologies and businesses. As IP tele-
phony gains market share, it is likely to have a dramatic impact on the
traditional, regulated voice service providers. Such developments have
provoked and will continue to provoke calls for voice over IP to be subject
to regulation akin to that in place for circuit-switched voice services or for
the regulatory regime to be modified in other ways to cover this new form
of telephony. As new IP- and Internet-based services emerge that in some
way resemble PSTN services that are currently regulated, there will be a
number of questions about whether these new services should be treated
in the same fashion. At the same time, in the face of competition, PSTN
operators may develop and offer new services in order to attract custom-
ers, raising questions about how those services will be accommodated
within the existing regulatory framework.
The position of the Federal Communications Commission has been to
keep the Internet free from unnecessary application of the existing regula-
tions on telecommunications, and many have argued that this hands-off
approach has been a significant factor in the Internet's explosive growth.l6
16See, for example, Jason Oxman. 1999. The FCC and the Unregulation of the Internet.
Office of Plans and Policy (OPP) Working Paper No. 31. Washington D.C.: OPP, Federal
TELEPHONY AS A CASE STUDY
171
Nonetheless, to the extent that IP telephony applications increase their
market share, it is reasonable to anticipate that pressure will come from
the other players (e.g., other telephony carriers and consumers) to in-
crease regulatory attention to Internet telephony.
The potential inconsistency between the assumptions underlying ex-
isting regulations and those that would be applicable to new forms of
telephony is well illustrated by one concern raised by local exchange
carriers that has been the subject of FCC regulatory attention: Would
local termination tariffs be applied to phone calls carried on the Internet?
Internet telephony services (or the ISPs that carry the data associated with
the phone call) do not, for example, have to pay local exchange carriers
for terminating calls when one of the callers is connecting to the Internet
via a modem running over a PSTN telephone line (as a conventional long-
distance provider is required to do when it hands off a call to a local
exchange carrier). Some of the advantages of using the Internet for voice
are thus amplified or in some instances even driven by tariff and regula-
tory artifacts that treat IP and public switched telephone networks rather
differently. Such advantages can be seen either as unfair to incumbents or
as appropriately reflecting the emerging, evolving nature of telephony.
A key question is whether IP telephony should be subject to common
carrier provisions. The implication of such status could be positive (e.g.,
no liability for the content it carries) or negative (e.g., the need to meet
certain standards of operational integrity that go beyond what is required
of an Internet, cable, or other noncommon carrier provider) for the IP
telephony provider. The Telecommunications Act of 1996 defines tele-
communications as the "transmission, between or among points specified
by the user, of information of the user's choosing, without change in the
form or content of the information as sent and received." A telecommuni-
cations carrier is defined as "any provider of telecommunications ser-
vices, except that such term does not include aggregators of telecommu-
nications services (as defined in section 226~." The act goes on to say that
"a telecommunications carrier shall be treated as a common carrier under
this Act only to the extent that it is engaged in providing telecommunica-
tions services, except that the Commission shall determine whether the
provision of fixed and mobile satellite service shall be treated as common
carriage.''l7 Finally, telecommunications is defined as "the offering of
telecommunications for a fee directly to the public, or to such classes of
users as to be effectively available directly to the public, regardless of the
facilities used."
Communications Commission, July. Available online at
72
THE INTERNET'S COMING OF AGE
These definitions immediately give rise to the question of whether IP
telephony providers or ISPs could fall under common carrier provisions.
Resolution of this sort of question will depend on how existing definitions
are applied to new technologies and services and, more fundamentally,
on whether it is deemed appropriate to apply existing definitions and
rules to these newly emerging services. For example, the act's definition
of telecommunications could be read to say that if the input is voice and
the output is intended to be the same voice, then regardless of the details
of how the voice signal is processed internally, voice communications
carried over IP would also be considered telecommunications.
One answer to the common carrier question was provided by the FCC
in its August 31, 1999, rulingl8 on the Communications Assistance for
Law Enforcement Act (CALEA) requirements (see Box 4.1~.19 CALEA
required telecommunications carriers to provide assistance for law en-
forcement in carrying out wiretaps. In its ruling, the FCC mandated that
cable television and IP services inasmuch as they provide telecommuni-
cations services and are a telecommunications carrier and thus a common
carrier, at least for the purpose of CALEA are subject to CALEA. The
FCC has given IP carriers until September 30, 2001, to comply.20 This
ruling opens the door for a broader interpretation and acceptance of IP
telephony as common carriage.
In addition to suggesting future directions with respect to the com-
mon carriage status of IP telephony, the effort to apply CALEA in this
area also suggests the sorts of societal expectations that surround tele-
phony. The advent of IP telephony raises the questions of whether and
how these expectations will be extended to the new technologies and
networks. One such expectation is the use of telephony for public safety.
Conventional wireline telephones provide enhanced 911 service, which
automatically provides public safety organizations with the address of a
911 caller. Cellular telephone service providers are being required to
provide analogous information for 911 calls that is, precise information
on the location of the caller.
18Federal Communications Commission (FCC). 1999. Second Report and Order in the Mat-
ter of Communications Assistancefor Law Enforcement Act. CC Docket No. 97-213, FCC 99-229.
Washington, D.C.: FCC, August 31. Available online at
TELEPHONY AS A CASE STUDY
173
Robustness and quality are also expected of telephony services, mani-
fested in state and federal regulatory attention to carrier performance.
The relative ease with which the Internet can offer voice services means
that many businesses will be in a position to offer voice services, probably
over a broad spectrum of service quality and reliability, from robust and
high quality to fragile, unreliable, and low quality. In addition, the ability
to overlay telephony on top of existing IP networks and the emergence of
74
THE INTERNET'S COMING OF AGE
mixed PSTN and IP network environments can be expected to raise new
questions about which parties are responsible for service quality. Busi-
ness imperatives and customer demands can be expected to address many
of these concerns, but consumers will still need to be clear about the
differences between service offerings (e.g., from a consumer protection
standpoint) and their acceptability from a performance standpoint (e.g.,
to meet public safety requirements).
There are also particular economic arrangements in the form of uni-
versal service charges and fees attached to telephony aimed at increasing
access to telephone service. Today, these are imposed only on conven-
tional telephony service. There are two views on this matter: Some
would like to see a level playing field across technologies and providers
and others are concerned about the potential for declining revenue from
these fees. (A related question is what form universal service takes with
respect to Internet service; see Chapter 5 for a discussion of the options.)
Another set of issues arises from inconsistencies between the new
technologies and architectures and the assumptions about architecture
that are embodied in the existing regulatory regime. Three such issues
are as follows:
· In Internet telephony, there are no meaningful distinctions between local
and interexchange carriers. The existing regulatory regime, in which local
and long-distance services are separated and in which customers select a
particular long-distance carrier to carry calls outside their local area, does
not map onto Internet telephony because the architectures of the two
networks are so different. In the PSTN, it is possible and indeed neces-
sary to specify the long-distance carrier over which a call is to be
switched. In contrast, the Internet architecture and protocols were not
designed to provide carrier selection capabilities like those in the PSTN.2~
Today, the only way in which the long-distance segment of a call could be
separated from the local segment would be a hand-off to voice services
over a PSN interexchange carrier. An insistence that local and long-
distance carriers be distinct entities could only be complied with by estab-
lishing requirements that run counter to the Internet's basic architecture.
· Owing to the separability of data transport and application service func-
tions, meeting some regulatory requirements that have been established for tele-
phony will be difficult or require different implementations. For example, it is
possible to construct a telephony service where the application servers
concern themselves only with providing directory services and call setup
However, under current regulations incumbent local exchanges that provide Internet
service must offer their subscribers a choice of long-distance IF network providers.
TELEPHONY AS A CASE STUDY
175
while the communications associated with the call depend only on the
public Internet. If a wiretap is ordered for a customer of an ISP, the
application service provider will have no means of allowing law enforce-
ment to access the content of the calls (although it could provide informa-
tion on the identity of the calling parties). The wiretap might require a
different technical approach, such as one that relies on accessing data
packets at their entry point to the Internet at the customer's ISP.
· Attempts to force the Internet to fit the existing regulatory model could
inhibit innovation by forcing modifications to the Internet's architecture. The
current regulatory apparatus is, for example, not set up to respond to a
world in which new telephony applications can be deployed simply by
having a third party distribute some new software and set up a few serv-
ers. Already, a number of telephony providers exist as overlays to the
public Internet, and it is reasonable to suppose that these will grow in
number and market share. One question these providers raise is whom to
hold responsible for meeting such mandates as provision of 911 service or
compliance with CALEA. There are two general ways in which this could
be done. The parties offering themselves as telephony providers could be
designated as the ones responsible for meeting the requirements. Or, the
requirements could be imposed on the Internet service providers. The
second alternative, while it might prove attractive because it makes the
ISPs responsible for determining who is and who is not providing voice
telephony, could have profound implications for the Internet. If ISPs
were required to ensure that voice traffic carried over their networks falls
under a particular set of rules, it would become necessary for them to
examine all the traffic over their network to screen out "unacceptable"
voice communications. Because telephony providers are free, in keeping
with the Internet's edge-based innovation model, to design their own
protocols for telephony, it can be impossible to reliably identify which
traffic is associated with telephony applications. Compliance with screen-
ing requirements might, in the end, make it necessary for ISPs to only
allow traffic of known, acceptable types to be transmitted, an outcome
that runs counter to the hourglass transparency of the Internet. This line
of reasoning is discussed in greater detail in Chapter 3.
LOOKING FORWARD:
THE INTERNET AND OTHER INDUSTRY SECTORS
The preceding sections illuminate, in the context of telephony, issues
of regulatory inconsistency, protection of incumbent companies, and re-
structuring of the service as perceived by the consumer. Technological
developments and deployment of the first generations of new services
176
THE INTERNET'S COMING OF AGE
make it apparent that the next decade or so will see similar collisions
between Internet-based businesses and other large industry sectors.
Looking beyond telephony, some of the easiest collisions to foresee
today are Internet-based distribution of music, which is exemplified by
the collision between the mp3 encoding protocol and traditional channels
of music distribution; the more general transmission of radio-like audio
content over the Internet; and, eventually, Internet-based television-like
services, which would collide with broadcasting. These applications are
not visions of the distant future. Internet-based music distribution is a
rapidly growing service, and while little entertainment video is transmit-
ted over the Internet, specialized applications such as continuing educa-
tion and training video are run over the Internet today. Many radio
stations are sending their content over the Internet simultaneously with
their over-the-airwaves broadcasts. As industry groups such as tradi-
tional network broadcasters, retail CD distribution chains, advertising
marketers, and large content-creation organizations find their markets
being nibbled at by Internet alternatives, they can be expected to react.
The result may be large transformations and dislocations in existing mar-
kets; the result may also be stressful for the Internet, its design principles,
and its service providers.
One predictable trend is the use of existing regulation by incumbents
to protect a legacy industry position. For example, in radio and television,
there are rules (about, for example, public access or political access) that
presume that space on the broadcast spectrum is scarce. Since capacity on
the Internet is not scarce and transmission does not require a federal
license, anyone can, in principle, generate and distribute content, and it is
not clear that there is a rationale for applying these rules to the Internet.
However, much as one comes across similar arguments with respect to
telephony, one can easily imagine calls for these rules to be imposed on
some forms of Internet content providers. (The argument would be that it
is unfair to impose different burdens on producers of similar sorts of
content that happen only to use different forms of distribution.) Another
area where existing practices and the capabilities afforded by the Internet
collide is copyright, where the assumptions underlying the current copy-
right regime are being stressed by the ability to make perfect copies of
digital works as well as by the ease with which they can be distributed
over the Internet. The technological, legal, economic, and social factors
surrounding copyright are too complex to analyze here; for more discus-
sion, see a recent CSTB study of these issues.22
22Computer Science and Telecommunications Board, National Research Council. 2000.
The Digital Dilemma. Washington, D.C.: National Academy Press.