Upload
milind-ravindranath
View
59
Download
0
Tags:
Embed Size (px)
Citation preview
IP Telephony in India: Cost, Pricing and Regulation
A Project report
Submitted to
����� ��� �� �
����� ������ ������ �
����� � �������
on Sept 5, 2000
in partial fulfillment of the requirements of the course
Infrastructure Development and Finance
by
���� �����
���� � ��
������ �� ������
����� ��
����������
�������� ��
��� �� ���� ���� �� ������������� �� ���� ���� �� ����������
����������������
2
Acknowledgements
We hereby acknowledge the contribution of the following people and resources to our project.
� Professors Rekha Jain, G.Raghuram and Sebastian Morris for laying a foundation in the
conceptual issues of infrastructure.
� Bhagabhati Maharana ,Infoscian and Ajith .A.Mascarenhas, Doctoral candidate at the
University of North Carolina ,U.S.A for providing an understanding of the technical aspects of
IP Telephony.
� The Internet Telephony Consortium at MIT, USA whose conferences, publications, and
research in the field of VoiP have aided us, especially in the field of regulatory and economic
issues.
-Amit Dhingra
Deepak Rajan
Manish K.Sarswat
Mudit Gera
S.Prashanth
3
Executive Summary
The advent of the Internet has set in motion a fundamental shift in the way telephony is viewed.
Traditional telephony or POTS (Plain Old Telephone Services) viewed voice as a continous
stream of electrical signals to be delivered uninterrupted (circuit switched) from one end to
another. Voice was the result of modulation and demodulation at either end. What the Internet has
achieved is to provide an alternate form of carrying these signals, albeit at a much cheaper and in
a more efficient form. This is through packetization of these data streams. This technology has its
drawbacks in that the Carrier views these packets as data packets and is liable to treat them as
such, which implies lost or delayed data packets, resulting in poor quality of voice. New
technologies are available today, which have brought down “Jitter”, “echo” and “latency” to a
more tolerable level. Worldwide, VoiP has already been commercialized as part of convergence
offerings.
This project aims to look at the applicability of IP telephony as a viable offering in India.
The delivery channels identified are ISP’s (who offer dial up connections), Cable (for its vast
penetration) and Phone to Phone (Incumbents like VSNL and MTNL). We feel that the latter two
are viable means of VoiP entry in India. Cable because of Multi media and convergence offerings
on broadband which would become possible in the near future and Phone to Phone, wherein we
feel that existing players would have an advantage as they would have to only replace the
backbone, and not the end instruments, which signify a captured market. The project takes a look
at regulatory issues, drawing parallels from regulatory ruling in the USA, which currently is in the
most advanced stage of implementation and usage of VoiP.The issue of pricing VoiP is a very
contentious one, as in the era of convergence, all manner of packets, i.e data, voice, video will
compete for access on the same network, unlike a circuit switched network. This throws up
questions of priority before pricing occurs.The model identified for pricing is based on the
concept of “Smart Markets” originally proposed by Hal Varian and Mackie-Mason in 1994.
This employs a bidding mechanism based on a “Vickrey Auction”. The costs of implementing
such a system in India have been worked out and it is found that it would cost on an average 25%
cheaper than existing POTS. We feel that in the context of Infrastructure development, this
technology with its price and technical advantage could drive higher accessibility and thus easier
flow of information across the country.
4
Table of Contents
What Is Packet Telephony?.......................................................................................................... 6
Carrier Applications............................................................................................................................. 6
Enterprise Applications ........................................................................................................................ 6
Voice Over Internet Quality of Service (QoS) Issues.................................................................... 7
QoS defined........................................................................................................................................... 7
Delay...................................................................................................................................................... 7
Accumulation Delay (Sometimes Called Algorithmic Delay) .............................................................. 8
Processing Delay ................................................................................................................................... 8
Network Delay ...................................................................................................................................... 8
Jitter ...................................................................................................................................................... 8
Lost-Packet Compensation................................................................................................................... 9
Echo Compensation ............................................................................................................................ 10
The Unique Demands of Voice Traffic............................................................................................... 10 Table 1. Characteristics of Packet and Circuit Switched Networks.................................................11
Engineering Around Delay and Packet Loss...................................................................................... 11
Is Overprovisioning of Bandwidth a Solution for QoS ? .................................................................. 12
MPLS- The Need for Traffic Engineering .........................................................................................12
Differentiated Services (DiffServ or DS) ............................................................................................ 13
IPv6 ..................................................................................................................................................... 14 Latest Developments on Ipv6...................................................................................................................15
Challenges of IP Telephony ................................................................................................................ 18
CABLE IP TELEPHONY.......................................................................................................... 20
Cable IP telephony- Architecture....................................................................................................... 20
Cable IP telephony - The evolution to come....................................................................................... 22
Long Distance Service for ISPs.................................................................................................. 27
Deploying VoIP in an ISP’s Network ................................................................................................. 28 Components of a H.323 Network............................................................................................................31
Service Description ............................................................................................................................. 32 How the Service Works—Call Processing............................................................................................32 Facilities embedded in the network:-......................................................................................................33
Regulation: Legal and Policy Issues.......................................................................................... 35
US Scenario......................................................................................................................................... 35
The Petition......................................................................................................................................... 36
The Proceedings.................................................................................................................................. 36
5
The Ruling........................................................................................................................................... 36
FCC’s Report to Congress.................................................................................................................. 37 Stance of European Commission.............................................................................................................38 Stance of other Government/Regulators................................................................................................40 Interoperability of IP Telephone..............................................................................................................40
INTERNET TELEPHONY IN INDIA....................................................................................... 42
MODES OF PROLIFERATION ....................................................................................................... 43 THE ISP SCENARIO IN INDIA............................................................................................................43
How would the future Indian IP Telephony market look like?......................................................... 47 Retail Segment.............................................................................................................................................47 Corporate Segment.....................................................................................................................................47
Current Legal and Policy Concerns In India.....................................................................................47
What lies ahead for the Incumbents?................................................................................................. 48
PRICING OF IP TELEPHONY........................................................................................ 50
Cost of IP Telephony in India............................................................................................................. 52 Key Assumptions........................................................................................................................................52
Results ................................................................................................................................................. 53 How cheap is IP Telephony compared to traditional telephony?.....................................................54 Major Hurdles..............................................................................................................................................54 Possible Answers........................................................................................................................................54
Exhibit 1: Cost estimates for connection for IP Telephony ............................................................... 55 VSNL Cost.......................................................................................................................................................55
Exhibit 2: Prevailing STD and ISD rates in India ............................................................................. 56
Exhibit 3: Comparison of prices of IP Telephony (PC to Phone) with Traditional Telephony prices (US prices for overseas telephony) ..................................................................................................... 57
Bibliography...............................................................................................................................61
6
What Is Packet Telephony? There is a lot of confusion regarding packet telephony in two specific areas. First, packet
telephony is not Internet telephones. Internet telephones are software packages sold
predominantly to place telephone calls over the Internet. They are generally awkward to use and
offer poor voice quality.
Packet or IP telephony is the simultaneous and joined delivery of voice and data communications
over a single, unified communications fabric based upon the Internet Protocol (IP). Packet
telephony traffic will be delivered within the enterprise over an organization’s intranet and
outside the enterprise initially over a circuit-switched fabric. Over time, as corporations develop
extranets with their trading partners and as those partners install interoperable packet telephony
systems, corporations will deliver packetized voice end-to-end outside of the enterprise.
������� �������� �In the carriers, packet telephony is emerging as a key bypass technology. A new class of carriers,
Internet telephony service providers (ITSPs), is building packet-based WAN networks to carry
voice traffic. Even some traditional long-distance carriers are experimenting with packet-based
WANs, primarily for service outside their regulated markets.
� �������� �������� �In the enterprise, packet telephony will emerge in applications where the value proposition can be
clearly articulated. This is likely to begin with specific applications in large organizations—
applications utilizing the joined delivery of data and voice over a single infrastructure. Examples
include next-generation call centers, new voice logging systems, and unified messaging: the
joined receipt of voice mail and e-mail. Because of its radical technological departure, packet
telephony will emerge at the fringe of organizations in value-added applications, not in the core
premises telephony fabric.
Much of packet telephony’s present deployment is for toll bypass over WANs. Organizations are
purchasing VOIP gateways to combine their phone and data infrastructures (or at least selected
links in their infrastructures) primarily to high-tariff countries.
Domestically, organizations with large branch office systems (such as banks) and a capillary data
infrastructure to reach those branches are also exploiting the WAN data network to carry voice. In
an odd reversal of the metaphor that sold T1 multiplexers for years (“voice pays for the circuit;
7
data rides for free”), these customers have already installed and justified their data networks and
are using spare capacity to carry intra-corporation voice traffic. Over time, as the business
justification shifts to increased employee productivity and effectiveness, packet telephony will
infiltrate the LAN fabric. Several events need to occur to effect this change. Much of the high
cost of installing a LAN-based telephony system today lies in the infrastructure components (as
opposed to telephone handsets or other client equipment).
Voice Over Internet Quality of Service (QoS) Issues
��� ���� ��Quality of service can be interpreted as the ability of a user of a specific application to obtain
service with predictable performance over some reasonable period of time that permits the
application to operate in an acceptable manner. Latency and dropped packets from congested
links make it extremely difficult to provide such a predictable performance. In general, delivering
QoS over the Internet depends upon two primary components: -
1. The use of an explicit bandwidth reservation tool for reserving network resources tailored to
specific application flows. The expectation for QoS to exist in parallel with a “best-effort
Internet” for those who are willing to pay a premium for more predictable service quality. Thus
it is imperative for network operators to utilize management tools like admission control to
restrict access to reserved and conformant traffic only.
2. The routers must maintain control mechanisms to enforce preferred access to network
resources for priority flows.
Some of the problems associated with VoIP are discussed below in greater detail..
����Delay causes two problems: echo and talker overlap. Echo is caused by the signal reflections of
the speaker's voice from the far-end telephone equipment back into the speaker's ear. Echo
becomes a significant problem when the round-trip delay becomes greater than 50 milliseconds.
As echo is perceived as a significant quality problem, voice-over-packet systems must address the
need for echo control and implement some means of echo cancellation.
Talker overlap (or the problem of one talker stepping on the other talker's speech) becomes
significant if the one-way delay becomes greater than 250 milliseconds. The end-to-end delay
budget is therefore the major constraint and driving requirement for reducing delay through a
packet network.
8
The following are sources of delay in an end-to-end, voice-over-packet call:
�������� ���� ���������� ���� ��������� �����This delay is caused by the need to collect a frame of voice samples to be processed by the voice
coder. It is related to the type of voice coder used and varies from a single sample time (.125
microseconds) to many milliseconds. A representative list of standard voice coders and their
frame times follows:
� G.726 adaptive differential pulse-code modulation (ADPCM) (16, 24, 32, 40 kbps)—
0.125 microseconds
� G.723.1 Multirate Coder (5.3, 6.3 kbps)—30 milliseconds
������� � ����This delay is caused by the actual process of encoding and collecting the encoded samples into a
packet for transmission over the packet network. The encoding delay is a function of both the
processor execution time and the type of algorithm used. Often, multiple voice-coder frames will
be collected in a single packet to reduce the packet network overhead. For example, three frames
of G.729 code words, equaling 30 milliseconds of speech, may be collected and packed into a
single packet.
������� ����This delay is caused by the physical medium and protocols used to transmit the voice data and by
the buffers used to remove packet jitter on the receive side. Network delay is a function of the
capacity of the links in the network and the processing that occurs as the packets transit the
network. The jitter buffers add delay, which is used to remove the packet-delay variation to which
each packet is subjected as it transits the packet network. This delay can be a significant part of
the overall delay, as packet-delay variations can be as high as 70 to 100 milliseconds in some
frame-relay and IP networks.
�����The delay problem is compounded by the need to remove jitter, a variable inter-packet timing
caused by the network a packet traverses. Removing jitter requires collecting packets and holding
them long enough to allow the slowest packets to arrive in time to be played in the correct
sequence. This causes additional delay.
The two conflicting goals of minimizing delay and removing jitter have engendered various
schemes to adapt the jitter buffer size to match the time-varying requirements of network jitter
removal. This adaptation has the explicit goal of minimizing the size and delay of the jitter buffer,
while at the same time preventing buffer underflow caused by jitter.
9
Two approaches to adapting the jitter buffer size are detailed below. The approach selected will
depend on the type of network the packets are traversing.
The first approach is to measure the variation of packet level in the jitter buffer over a period of
time and incrementally adapt the buffer size to match the calculated jitter. This approach works
best with networks that provide a consistent jitter performance over time, such as ATM networks.
The second approach is to count the number of packets that arrive late and create a ratio of these
packets to the number of packets that are successfully processed. This ratio is then used to adjust
the jitter buffer to target a predetermined, allowable late-packet ratio. This approach works best
with the networks with highly variable packet-interarrival intervals—such as IP networks.
In addition to the techniques described, the network must be configured and managed to provide
minimal delay and jitter, enabling a consistent QoS.
!���"����� ����� ����� Lost packets can be an even more severe problem, depending on the type of packet network that
is being used. Because IP networks do not guarantee service, they will usually exhibit a much
higher incidence of lost voice packets than ATM networks. In current IP networks, all voice
frames are treated like data. Under peak loads and congestion, voice frames will be dropped
equally with data frames. The data frames, however, are not time sensitive, and dropped packets
can be appropriately corrected through the process of retransmission. Lost voice packets,
however, cannot be dealt with in this manner.
Some schemes used by voice-over-packet software to address the problem of lost frames are as
follows:
� interpolate for lost speech packets by replaying the last packet received during the interval
when the lost packet was supposed to be played out; this scheme is a simple method that
fills the time between noncontiguous speech frames; it works well when the incidence of
lost frames is infrequent; it does not work well if there are a number of lost packets in a
row or a burst of lost packets.
� send redundant information at the expense of bandwidth utilization; this basic approach
replicates and sends the nth packet of voice information along with the (n+1)th packet;
this method has the advantage of being able to correct for the lost packet exactly;
however, this approach uses more bandwidth and also creates greater delay
10
� use a hybrid approach with a much lower bandwidth voice coder to provide redundant
information carried along in the (n+1)th packet; this reduces the problem of the extra
bandwidth required but fails to solve the problem of delay
��� ����� ����� Echo in a telephone network is caused by signal reflections generated by the hybrid circuit that
converts between a four-wire circuit (a separate transmit and receive pair) and a two-wire circuit
(a single transmit and receive pair). These reflections of the speaker's voice are heard in the
speaker's ear. Echo is present even in a conventional circuit-switched telephone network.
However, it is acceptable because the round-trip delays through the network are smaller than 50
milliseconds and the echo is masked by the normal side tone every telephone generates.
Echo becomes a problem in voice-over-packet networks because the round-trip delay through the
network is almost always greater than 50 milliseconds. Thus, echo-cancellation techniques are
always used. ITU standard G.165 defines performance requirements that are currently required
for echo cancellers. The ITU is defining much more stringent performance requirements in the
G.IEC specification.
Echo is generated toward the packet network from the telephone network. The echo canceller
compares the voice data received from the packet network with voice data being transmitted to
the packet network. The echo from the telephone network hybrid is removed by a digital filter on
the transmit path into the packet network.
Without appropriate design, this can wreak havoc on a telephone conversation. A number of
techniques have been developed to address this problem. One technique is to use jitter buffers to
smooth out the ebb and flow of packets. However, jitter buffers, which store a string of packets,
introduce additional absolute delay.
#�� $ �%�� ���� �� �� &��� #�����The legacy PSTN was engineered to optimize the transport of predictable analog voice traffic. Its
centralized, circuit switched nature is not suitable for meeting the demands of large volumes of
data traffic. Instead, voice is migrating onto distributed, packet switched networks such as those
running IP, Asynchronous Transfer Mode (ATM), and Frame Relay technologies. Circuit and
packet switched networks have distinct differences, as shown in Table 1.
11
Table 1. Characteristics of Packet and Circuit Switched Networks
Packet Networks Circuit Networks
Architecture Distributed Centralized
Transport Packet/statistical multiplexing Circuit/time-division multiplexing (TDM)
Bandwidth Broadband Narrowband
Optimum Application(s) Multi-service Voice
� �� ���� � ���� � ���� � � ����� !���Voice quality is significantly impacted by one-way network delays of anywhere between 100 and
200 milliseconds (ms), depending on the tolerance level of the individual. Contributors to delay
include normal processing in a network, the distance between two communicating points, network
congestion, and packet loss and retransmission delays. To successfully deploy toll-quality voice
services over packet infrastructures, service providers must engineer their networks in a way that
ensures that latency does not exceed the 200 ms end-to-end, one-way maximum. To
accommodate a wider scope of users, keeping latency below 100 ms is preferable. Packet loss
should remain under 3 percent, with 1 percent packet discard preferable.
Consistently achieving these metrics requires equipment with integrated Quality of Service (QoS)
mechanisms for prioritizing, queuing, and reserving bandwidth for delay-sensitive traffic. Among
these mechanisms are standards-based options such as DiffServ, Int-Serve, Resource Reservation
Protocol (RSVP), and Multiprotocol Label Switching (MPLS). These technologies enable service
providers to enter into service level agreements (SLAs) with their customers to guarantee certain
network metrics so that users are assured that their applications will perform well over the
network. Some of them are discussed below in the study.
Simply over-provisioning bandwidth to the point of eliminating any network congestion is
another approach to controlling these metrics. The expense associated with this alternative,
however, can negate the cost savings of deploying an integrated IP infrastructure. In addition, the
extra bandwidth will only suffice until traffic volumes increase. Traffic engineering and
performance monitoring provide a more economical option for controlling network performance.
12
'� ()�����)���� � � �� *� ������ � ������ ��� ���+
The simplest and the most obvious solution to provide better service quality to meet the complex
demands is to deploy ample bandwidth in the core to keep ahead of the demand curve. This
motivation has necessitated the constant upgrading of backbone fiber links by core transit
providers. However, this strategy of deploying an increased number of higher data rate channels
over fibre has not satisfied the hunger for bandwidth- in fact the demand has grown so as to
consume the bandwidth available.
Overbuilding networks is a highly capital intensive solution that may not continue to scale in a
world where the cost of bandwidth is declining more dramatically than the network transmission
costs. Before companies can migrate to critical business applications on the internet, a new set of
traffic management tools is required to align the bandwidth requirements of more varied traffic
types with the physical resources that the network can provide. These requirements create the
underpinnings for the emergence of traffic engineering and the QoS. While overprovisioning can
relieve congested traffic arteries, it delivers no tightly encompassing mechanisms for guaranteed
service level agreements (SLAs) encompassing delay intolerant applications.
,�!�" #�� ���� ��� #����� � �� ���� �
The IP was created as a connectionless network layer protocol that makes no attempt to
discriminate the various application types. IP uses traditional interior gateway protocols to
advertise and build a database of all active links within a routing domain. Successful operation of
these networks depends upon the same distributed network state information being disseminated
and consistently maintained by all routers within the same autonomous system. Each router uses
the same global state information to independently develop its own forwarding table using the
shortest path constraint –based metrics. As a result, the traffic is concentrated on a small number of
optimized data paths to the detriment of other links, which frequently remain underutilized.
To accommodate highly interactive application flows with low delay and packet loss thresholds,
there is a clear need to more efficiently utilize the available network topology. The process
13
whereby this is accomplished is known as traffic engineering and Multi Protocol Label Switching
(MPLS) provides these capabilities.
MPLS traffic engineering attempts to correct the inefficiencies of typical datagram routing
protocols by more evenly spreading the flow of traffic across all available resources.
Reengineering a conventional datagram network based solely on Layer 3 cost-based metrics can be
both expensive and inefficient as network re-convergence times are higher and it means moving all
data flowing across a link to an alternate path. In an MPLS traffic engineered path, when a more
desired route becomes available, some labels associated with certain traffic classes may be
assigned the optimal path while delay-intolerant service classes may remain behind on the original
link.
Through more precise balancing of various traffic engineering mechanisms, MPLS provides an
extensive array of tools for more precise balancing of flows of different size and application
priority across the most lightly loaded network links. The goal of traffic engineering is to increase
through put across a network while concurrently decreasing congestion. As a result, the preferred
paths in the new networks may not be synonymous with the shortest cost paths. Ina coat
competitive market for bandwidth, MPLS provides an effective tool for increased network
utilization and yields economies of scale and relative cost advantages for service provider.
������� ������ ���)��� ��������) �� ���Differentiated Services (DiffServ, or DS) is a protocol for specifying and controlling network
traffic by class so that certain types of traffic get precedence - for example, voice traffic, which
requires a relatively uninterrupted flow of data, might get precedence over other kinds of traffic.
Differentiated Services is the most advanced method for managing traffic in terms of what is
called Class of Service (CoS). Unlike the earlier mechanisms of 802.1p tagging and Type of
Service (ToS), Differentiated Services avoids simple priority tagging and depends on more
complex policy or rule statements to determine how to forward a given network packet. An
analogy is made to travel services, in which a person can choose among different modes of travel -
train, bus, airplane - degree of comfort, the number of stops on the route, standby status, the time
of day or period of year for the trip, and so forth. For a given set of packet travel rules, a packet is
given one of 64 possible forwarding behaviors – known as per hop behaviors (PHBs). A six-bit
field, known as the Differentiated Services Code Point (DSCP), in the Internet Protocol (IP) header
specifies the per hop behavior for a given flow of packets.
14
Differentiated Services and the Class of Service approach provide a way to control traffic that is
both more flexible and more scalable than the Quality of Service approach.
'�)-
IPv6 (Internet Protocol Version 6) is the latest level of the Internet Protocol (Internet Protocol)
and is now included as part of IP support in many products including the major computer operating
system. IPv6 has also been called "IPng" (IP Next Generation). Formally, IPv6 is a set of
specifications from the Internet Engineering Task Force (IETF). IPv6 was designed as an
evolutionary set of improvements to the current IP Version 4. Network host and intermediate node
with either IPv4 or IPv6 can handle packet formatted for either level of the Internet Protocol. Users
and service providers can update to IPv6 independently without having to coordinate with each
other.
The most obvious improvement in IPv6 over the IPv4 is that IP addresses are lengthened from 32
bits to 128 bits. This extension anticipates considerable future growth of the Internet and provides
relief for what was perceived as an impending shortage of network addresses.
IPv6 describes rules for three types of addressing: uni-cast (one host to one other host), anycast
(one host to the nearest of multiple hosts), and multi-cast (one host to multiple hosts). Additional
advantages of IPv6 are:
� Options are specified in an extension to the header that is examined only at the destination,
thus speeding up overall network performance.
� The introduction of an "anycast" address provides the possibility of sending a message to
the nearest of several possible gateway hosts with the idea that any one of them can
manage the forwarding of the packet to others. Anycast messages can be used to update
routing tables along the line.
� Packets can be identified as belonging to a particular "flow" so that packets that are part of
a multimedia presentation that needs to arrive in "real time" can be provided a higher
quality-of-service relative to other customers.
� The IPv6 header now includes extensions that allow a packet to specify a mechanism for
authenticating its origin, for ensuring data integrity, and for ensuring privacy.
15
Latest Developments on Ipv61
WorldCom Inc. is testing IPv6 within its own VBNS (very-high-performance Backbone Network
Service). The company runs a nationwide native IPv6-over-asynchronous transfer mode section on
the VBNS at speeds of 155M bps. Right now, the driving force toward IPv6 is increased
addressing space, especially as Internet-enabled devices become prevalent. Under the IPv4
specification for Internet traffic, which uses a 32-bit addressing scheme, it is to run out of IP
addresses eventually. The 128-bit addressing scheme of IPv6 is enough to dedicate many
thousands of IP addresses for every square inch of the Earth’s surface.
Sprint and the '6bone' network
Sprint Corp., for its part, hopes to get an early jump on the move to IPv6 through experiments with
the "6bone" network -- a network put in place by industry groups to run early versions of IP v6.
The company operates its portion of the 6bone network running IPv6 traffic exclusively as a
"tunnel" within its Internet backbone. Sprint currently provides IPv6 connections to about 70
research, government, academic and corporate entities.
Sprint is also looking to IPv6 as the vehicle for far higher levels of service security and
performance, he said. For instance, IPv6 was engineered to include built-in support of QOS,
DiffServ (Differentiated Services) and IPSec (IP Security). IPv4 includes none of those functions
as a uniform part of the standard.
Despite the work being done with the specification, commercial availability of IPv6 software and
services isn’t expected until next year at the earliest, after networking companies such as Cisco
Systems Inc. begin implementing the technology in core products.
Evolution of Market in the United States
The market for Internet Telephony was largely created by VocalTec's Internet Phone,
introduced in February 1995. These first-generation Internet Telephony products were
characterized by the fact that users on both ends of the conversation are required to have Internet
connected computers with compatible software to convert voice into data packets. To locate people
on the Internet, most applications use an scaled down LDAP (lightweight directory access
protocol) directory service that allows an e-mail address to be mapped to the users current IP
1 www.zdnet.com, July, 2000
16
address -- user must logon to the LDAP directory each time they connect to the Internet. Since that
time, there has been a significant increase in offerings of Internet Telephony software from many
vendors. VocalTec claims there have been over 600,000 downloads of its test software since its
introduction. While these applications vary in sound quality, using 28.8 Kbps modems on both
ends can provide sound quality comparable to a regular phone call most of the time, international
calls tend to have lower quality.
According to the research group Frost and Sullivan, the concept of IP Telephony has evolved
from the initial use popularized by VocalTec into the following five areas:
1. Voicemail: Non-real-time audio communication where one person sends a voice message
to another person.
2. Fax: Near real-time and store-and-forward data communication between two users.
3. Voice telephony: Real-time audio communication between two or more users.
4. Desktop videoconferencing: Real-time audio-visual communication between two or more
users, where each user can see the others on a computer screen.
5. Application sharing and document sharing: The sharing of software applications and
documents, in real time, by at least two users (application sharing); the sharing of
documents, in real time, by at least two users (document sharing). Document sharing is
different from application sharing in that no user can take control of someone else's
application. Every participant can view and modify the document using his or her own
application. (Network Computing, NC-C, NC-S, NetPC)
As described above, the first three areas have been traditionally functions of voice networks
and were services performed by PBX type telephone switches. The last two areas have been
applications promoted as tools used by GroupWare electronic collaboration and real-time
conferencing applications. Groupware and group collaboration products tend include IP Telephony
features. Netscape Communicator and Microsoft's NetMeeting are freely available on the Internet;
as end-users start experimenting with the applications, the Internet will see an increase in the use
of IP Telephony. The FCC has decided that these first-generation PC based IP Telephony
applications are enhanced service products, not subject to regulation.
In 1996 VocalTec introduced the second generation of IP telephony products by marketing a
gateways product that would allow for the translation of Internet domain IP addresses with voice
network telephone numbers. The gateway products allow for a call originating on an Internet
connected PC to be terminated at an analog based telephone; typically an integration of an Internet
17
based LDAP directory with the telephone system's SS72 call processing protocol. In 1997, the
International Multimedia Teleconferencing Consortium adopted the H.3233 conferencing standard
as the standard for Internet telephony. These second-generation products tend to revolve around
PBX telephone systems that operate as gateways to the Internet with the Virtual IP network being
treated as another long-distance carrier in terms of least-cost call routing.
Gateways consist of three parts:
1. Codec -- software that typically runs on a dedicated PC and handles the packetization,
compression, and decompression of voice calls.
2. Line Interface -- interconnects with either a company's PBX (Private Branch Exchange) or
the PSTN (public switched telephone network.)
3. Session Management -- manages the connection between gateways
In a corporate environment, when the user dials a number in another city, a setup message is sent
to the local PBX, which in turn sends the message to the gateway. The local gateway consults a
lookup table to locate the IP address of the remote PBX's gateway. The two gateways establish a
session and the remote PBX is requested to complete the call. As the remote telephone rings, the
gateways run voice traffic through the codec and sends voice packets to the appropriate gateway.
Examples of Internet telephony gateway web sites:
Lucent Technologies (http://www.lucent.com/) Internet Telephony Server
RADvision Ltd. (http://www.radvision.com/l2w323.html) L2W-232
Vocaltec Ltd. (http://www.vocaltec.com/) Vocaltec Telephony Gateway
One of the major problems of a public Internet telephony system is how to share costs
between users and providers. Telecommunication carriers such as MCI Worldcom can purchase
unbundled elements from Local Exchange Carriers and resell them to ISPs and other large
customers through their UUNet service.
Third-generation IP Telephony products will provide Gatekeeper functionality that allows
for transparent translation between IP addresses and International Direct Distance Dial (IDDD)
2 Traditional circuit-switched voice networks currently that typically use SS7 switching protocols. 3 H.323 is a comprehensive International Telecommunications Union (ITU) standard for multimedia communications (voice, video, and data) over connectionless networks that do not provide a guaranteed quality of service, such as IP-based networks and the Internet. It provides for call control, multimedia management, and bandwidth management for point-to-point and multipoint conferences
18
telephone numbers, as well as a settlement system that will allow revenues to be split between
sending and receiving gateways. A computer connected to the Internet will be able to receive a
voice telephone call, without a second telephone line and without the end-user disconnecting from
the ISP. Because of efficiencies in the data network over the voice network, it is estimated that
ISPs would be able to earn twice the profits from a voice call as from a data call and still be
competitive with long-distance carriers.
Several companies are providing gateway and gatekeeper services between. The
Public Switched Network and the Internet and are: -
Delta Three (http://www.deltathree.com/)
Net2phone (http://www.net2phone.com/english/)
RSL Communications - Delta Three service is available in 16 countries
(http://www.rslcom.com/)
���� ��� �� '� #����� �Since IP packets carrying voice are treated just like IP packets carrying any other type of data, they
are subjected to delays, loss, and retransmissions. This is especially true when the network is
congested. The quality of service becomes very important issue. Losing every other words of the
phone call can make the call essentially worthless. IP telephony is facing the following challenges:
� Unpredictable service quality, which relates to quality of service and reliability. Real-time
applications set high requirements on the reliability and quality of service capabilities of IP
networks. Protocols and techniques to ensure this must be developed. Until these
techniques are widely deployed and supported by most networks, over-provisioning or
proprietary methods in private IP networks remain the only way to ensure the required
QoS.
� Datacom and telecom convergence related complex system integration, Network
Management Systems (NMS) integration, Customer Care and Billing (CCB) systems
integration, and diversity in the marketplace. IP telephony equipment consist of new
network elements that need to be integrated into the corporate, and teleoperator's or service
provider's network. Both physical and logical integration to the other network elements are
required, as well as integration to the vital operation support systems such as maintenance,
provisioning, and billing systems.
19
� Lack of interoperability because a single waterproof standard does not exist. There are
several competing or partially overlapping standard proposals. Current IP telephony
standards only ensure interoperability within a single IP telephony subnetwork. The
communication between gateways or gatekeepers from different vendors remains to be
standardized.
� Regulatory development will have a major impact on IP telephony. In most countries IP
telephony is still unregulated but the regulatory authorities are monitoring the situation
closely.
� Inertia in the legacy networks, large investments tied in legacy technologies, and people are
accustomed to the old services. There is inertia in the traditional telecom services.
20
CABLE IP TELEPHONY
��.� '� ������ �" ����������
In cable networks, both circuit-switched and IP technologies are overlaid onto a hybrid fiber/coax
(HFC) network, which uses optical fiber for signal distribution along trunks or between cable
headends. Fiber's resistance to noise and low signal attenuation drove its acceptance as a preferred
trunk medium in cable networks. As fiber became more prevalent, its costs decreased, which
helped extend it further into the cable network.
Today, a common configuration for larger service providers is a star-star-tree topology. In this
design, fiber extends from the headend to hubs in the first star layer, and then from hubs to fiber
nodes in a second star layer. The final layer begins with optical to electrical conversion at the fiber
nodes and continues with sets of coax tree branches from the fiber node to the subscriber.
Adding circuit-switched telephony to an HFC network requires three network elements: a
telephony switch, a network interface unit (NIU) and a host digital terminal (HDT). Their
placement in an HFC cable network is shown in Figure 1.
HDT: Host digital Terminal
HFC: Hybrid Fiber Coax
NIU : Network Interface Unit
Information Source: Cable Datacom News, www.kineticstrategies.com
NIU
Fiber Node
Coax
Video Signal Processing at headend
Optical-Electric Conversion
HDT
Telephony Switch
Public Network
Fiber
Fiber
Figure 1: HFC switched telephony Network
21
The functions of a telephony switch can be grouped into three categories: call processing, call
routing and feature provisioning. Call processing and call routing set up a path through the public
network between parties when a call is initiated. Enhanced capabilities such as call screening are
provided from the digital switch through the HFC telephony network. Associating these feature
capabilities with an originating line is an important part of the switch's function. Switch software
can also provide call forwarding, call transfer, call waiting and call conferencing.
In telephony, the NIU is the demarcation point between subscriber-owned equipment and
equipment owned and maintained by the service provider. The NIU has outgrown its traditional
role as a passive termination point. Although it still terminates the cable company's coax plant at
the customer premises, it must also provide:
� Twisted pair termination for telephony.
� Analog-to-digital conversion and vice versa for voice telephony.
� Packetization of digital information.
� RF modem.
� Diagnostics.
� Dial tone and ring generation.
As the interface between a cable distribution system and the telephony switch, the HDT acts as a
digital multiplexer. It provides T-1/E-1 links to the telephony switch at 1.544 Mb/s and accepts 64
kb/s digital signals from lines on the subscriber side, usually in a T-1/E-1 format.
Most vendors have designed the HDTs with an open interface to the telephony switch, allowing
the service provider to choose different vendors for the switch and the HDT. On the subscriber
side, however, the connection to the NIU is proprietary, requiring the cable operator to purchase
both the HDTs and NIUs from the same vendor. Having an open interface to the switch also
enables a cable operator to obtain telephony switching from another company through alliances or
leasing agreements, so the operator doesn't need its own digital switch in the early stages of
telephony offerings.
22
��.� '� ������ � " #�� �)����� �� ���
Certainly, the existence of these technical issues does not mean that IP telephony will never reach
dominance or that operators shouldn't implement it in the short term. To the contrary, visionary
service providers can develop an IP telephony strategy that applies packet technology in parts of
the network when and where it makes sense.
One solution is to creatively locate gateways in an operator's network. For example, a gateway at
the subscriber's location could be part of a long-term strategic deployment, based on the evolving
standards previously discussed. On the other hand, gateways at the network side of a telephony
switch make sense now.
Such architecture enables a service provider to offer cost- and quality-based service options, using
the switch to choose routes through either the public network or the Internet. As the feature servers
are developed and installed, service providers could add gateway functionality at either the HDT
or the NIU. The result will be an integrated IP/circuit-switched telephony architecture (Figure 2).
Gateway Gateway
Terminals
* Policy Based Routers
Information Source: Cable Datacom News, www.kineticstrategies.com
Gatekeeper
Gateway Gateway
Gatekeeper
* *MCU HDT
Digital Switch
Digital Switch
HDT MCU
Public Network
INTERNET
Public DataNetwork
1 1
NIU NIU
HFC HFC
POTS VOIP TV POTS VOIP TV
Figure 2 : An integrated IP/circuit-switched telephony architecture
23
In fact, because the NIU can be remotely provisioned, vendors will likely build gateway
functionality into the NIU. This will allow operators to select whether a subscriber line is
provisioned to route via the IP network or the public network without a truck roll dispatch. This
flexibility will allow cable service providers to offer circuit-switched service today and voice over
IP when the supporting infrastructure is in place. Without a doubt, this evolutionary journey will
bring creative and unique solutions to the networks of tomorrow.
Standards and their evolution
There are predominantly two cable modem standards that operate in the cable industry. DOCSIS,
which is the standard in North America and in other International markets, and DVB/DAVIC
EuroModem, which is an emerging standard in Europe .Of these standards, DOCSIS is the
standard used by a majority of cable modem vendors worldwide.
The North American cable industry developed the DOCSIS standard to create a competitive
consumer market for cable modem equipment. Seeking to capitalize on the most obvious service
opportunity -- delivering high-speed Internet access -- the DOCSIS 1.0 standard was designed as a
cheap consumer Web-surfing platform. While well-suited for its intended application, DOCSIS 1.0
does not provide all of the QoS and latency controls required to offer toll-quality IP voice services.
A team at AT&T Labs identified three key items that must be added to DOCSIS 1.0 to
support toll-quality telephone calls: upstream packet fragmentation and reassembly techniques,
support for a national clock, and an advanced isochronous scheduling system.
Because DOCSIS products employ an asymmetric architecture, offering 27 Mbps of
downstream capacity and typically less than 1 Mbps upstream, packet fragmentation is required to
avoid upstream congestion that impacts call quality. Specifically, the largest Ethernet packet size is
1500 bytes. Thus, sending this full-size packet upstream over a 768-Kbps cable modem would take
about 15 milliseconds, straining the delay budget for a packet voice call. Using fragmentation
techniques, these large data packets are broken into smaller ones to prevent unacceptable
transmission delays.
24
The second item, a national clock, is necessary to properly synchronize transmissions between
cable modems on the network.
The final enhancement is adding a high-quality isochronous scheduler to headend-based DOCSIS
cable modem termination system (CMTS) equipment. Because the DOCSIS 1.0 standard was
designed as a consumer Internet access platform, system latency can run in the 50 to 70
millisecond range. While this window is fine for Web surfing, it seriously impacts packet
telephone call quality. To support packet telephony, CMTS vendors must offer an isochronous
scheduling solution closer to a two-millisecond time scale.
In the proposed PacketCable architecture, a range of DOCSIS 1.1-based client devices will
support IP telephone connections, including cable modems, digital set-tops and media terminal
adapters (MTAs), standalone devices that link telephone handsets to the cable data network. All of
these devices can be served in the same cable spectrum by a single DOCSIS CMTS.
Vendors expect that the addition of IP telephony support will only increase the cost of a DOCSIS
1.1 cable modem by 20 to 30 percent. Thus, an integrated cable modem and PacketCable MTA
could be priced as low as $250. Building DOCSIS 1.1 headend and client products with IP
telephony support is not easy. But an even greater challenge, according to cable operators and
vendors, is efficiently provisioning and managing the devices once they are installed on the
network. Additionally, engineering disparate local cable data systems and backbone networks to
offer high end-to-end IP voice quality is not trivial. This means voice packets need to be specially
identified and given priority by the CMTS, routers and switches as they traverse the network.
As a starting point, many cable operators favor initially deploying IP telephony merely as a local-
loop bypass service. In this scenario, voice packets would be transferred directly from the CMTS
to an IP telephony gateway, and then onto the public switched telephone network (PSTN). This
would enable cable IP telephony users to place and receive calls without using the incumbent local
exchange carrier (ILEC).
The ultimate goal of many MSOs is to also offer long-distance IP telephony over their
packet backbone networks. For example, a residential cable IP telephony customer served by
Comcast in Philadelphia might call another cable IP telephony customer served by Time Warner in
Los Angeles. The packet calls could be carried nationwide at very low cost without ever touching
25
a telephone company network. MSOs are currently evaluating options to enter into backbone
interconnection arrangements that would make such a solution viable.
Using IP, cable operators hope to create an integrated multi-service communications platform that
operates on a lower cost structure than existing circuit-switched alternatives, enabling aggressive
service price discounting without sacrificing margins. Besides undercutting competitors, MSOs
hope the flexibility of IP networks will allow them to deliver a host of unique value-added
features, such as integrated voice mail and e-mail messaging and the real-time provisioning of
additional phone lines without rewiring a home
Players like AT&T Corp., after deciding to back a fast-track approach to cable industry protocols
supporting IP telephony over cable networks, late last year signed off on a voice services
expansion plan for their own cable systems that puts off the transition to IP well into next year, if
not later. The reason they gave in making their decision public was that routers just weren’t ready
for the job.4
Where the total per-customer cost of provisioning voice over a system provided by
Arris, a joint venture between Antec Corp. and Nortel Networks , has been about $890 at the
outset of the company's first voice service deployment in California, it is anticipated that the
cost will fall to about $590 by the quarter of this year. This is predicted in the context of new
players, including Lucent Technologies Inc and Motorola Inc.
It is believed that one key reason that the IP solution will eventually prove more cost
effective is that the premises-mounted network interface units that support IP voice over cable
will use the cable modem technology that is also used for high-speed data, thereby eliminating the
cost of the separate modem that's required for current voice customers who want high-speed data.
One of the cable companies taking issue with AT&T as to how ready the IP platform is
for commercial deployments is Comcast Corp. has begun testing VoIP services in New Jersey.,
using the PathStar access router/server supplied by Lucent. Comcast's trial is taking advantage of
4 www.soundingboardmag.com, June 2000
26
the PacketCable specifications to deliver multiple lines of toll-quality service, complete with all
the features customers get from the telephone companies.
DiffServ and MPLS have emerged as the key to QoS in IP routing, with DiffServ used to
identify traffic flows within pre-set classes so that bit streams are aggregated and treated
according to the class specs on a per-hop basis from router to router. MPLS is applied on a
systems-wide basis using labels assigned at Layer 3 to create label-switched paths, thereby
affording carriers a means of maximizing traffic flow efficiency across the network in accordance
with the assigned priorities.
Further buttressing the case for router-based telephony is the emergence of terabit routers,
which are designed to aggregate all forms of IP traffic across the backbone at speeds that are high
enough to allow carriers to fully exploit the carrying capacity of optical networks.
Avici is supplying terabit routers for testing by various carriers, including Deutsche
Telekom AG, GST Telecommunications Inc. and MCI WorldCom Inc. with plans to begin
delivering product for commercial deployment in the second quarter. The system is designed to
scale from single modules operating at 2.5gbps to an array of multiple modules seamlessly
interconnected to deliver aggregate throughput of up to 5.6 terabits per second.
Terabit routing will also play a fundamental role in expanding the efficiency of the
cable industry's play in IP telephony. Once the zone-to-zone specifications for cable IP telephony
are established in version 1.1 of the PacketCable standard, cable companies will be able to
interoperate with each other's IP voice services directly across IP backbones, avoiding the
intervening steps of converting signals to ATM or switched circuit formats.
27
Long Distance Service for ISPs
Increasing competition and worldwide deregulation are opening up new opportunities for Internet
service providers (ISPs) to enter the lucrative voice market. Killen and Associates has estimated the
worldwide voice over IP (VoIP) long distance services market to be worth $9.4 billion by 2002. ISPs
can leverage their existing data infrastructure and subscriber bases to deliver carrier-class long distance
voice services over low-cost Internet Protocol (IP) networks. With minimal upgrades to their existing
IP networks, ISPs can carry voice traffic over packet networks, opening the door to a variety of new
services. ISPs can use a voice-enabled Cisco access server as the VoIP gateway to offer carrier-class
domestic and international phone calls and real-time fax transmissions. Subscribers can make long
distance calls from home or office using their regular telephone or fax machine, or they can call from
other locations by entering an account number or password.
An ISP can gain the following advantages by offering long distance service:
• Provide a new service beyond Internet access.
• Increase revenue from existing points of presence.
• Expand their customer base.
• Bundle voice and data services for greater differentiation.
• Leverage lower-cost IP infrastructure with voice compression and silence suppression to offer
various competitively priced voice services.
After the IP infrastructure has been enabled for voice, ISPs can begin to offer value-added services
such as voice mail, unified messaging, Internet call waiting (alerting users to incoming voice calls
while on line) and virtual second line (ability to make and receive voice calls from the user’s PC).
These services can greatly increase the ISP’s revenue streams from Internet access subscribers.
The specific service described is a retail voice and fax offering to an ISP’s residential and business
subscribers.
Off-net calls are handed off to a traditional long distance provider at discounted rates. Billing is
usage based. Many other service variations are possible. ISPs can handle off-net traffic by
exchanging traffic with other ISPs by partnering, joining a consortium, or going through a
settlements company. They can offer wholesale VoIP services to other carriers. Managed VoIP
service for multi-location enterprises is yet another service model.
28
������ � &�'� � � '��/� �������ISPs are well positioned to deploy VoIP because of their existing IP infrastructure. The equipment
and software needed to implement VoIP can be added incrementally. ISPs already have Internet
POPs that connect to local exchange carrier (LEC) or public telecommunications operator (PTO)
central offices.
Figures 3 (a) & (b) show before and after views of a POP being equipped for long distance
service.
ISPs can add individual voice-enabled access server gateways equipped with additional T1 or E1
interfaces to the PSTN. Additional WAN bandwidth may be needed to support the extra voice
traffic. One or more gatekeepers are added to serve multiple gateways. Existing RADIUS servers
can be used for AAA as well as existing routers and Ethernet switches located in their POPs. The
ISP may need to add a billing server for voice services.
With this equipment in place, the ISP can either place connections to long distance carriers for off-
net calls at local POPs or choose to centralize such connections at a single POP location. Deciding
factors include traffic expectations and the geographic distribution of POPs. It may be more cost-
effective to choose one high-bandwidth connection in order to get better bulk rates from the long
distance provider. To expand its service coverage, an individual ISP might choose to partner with
ISPs in other regions or join a consortium to provide widespread coverage. Settlement firms allow
ISPs to provide national coverage by exchanging voice traffic with other ISPs. Finally, the ISP can
use a centralized service node equipped with programmable switches to add support for such
value-added services as debit cards and calling cards.
29
Radius Server
������ Servers
WAN PSTN
Billin g Servers
Figure 3 (a) –Dial POP , before
30
Voice Data
Billin g Servers Radius Server
WAN
Router
PSTN
H-323 Gateway H-323 Gatekeeper
Figure 3 (b): After Deployment of H-323 for VoiP in the long distance Network
31
Components of a H.323 Network The H.323 is an ITU standard that provides a common foundation of data, audio and video and
communications across IP networks. The elements required in a H.323 network are as follows:-
1. H.323 gateway
2. H.323 gatekeeper
3. RADIUS server
4. Billing server
1.H.323 gateway
The H.323 gateway is the gateway between the PSTN and the H.323 packet-switched network. It
provides standard interfaces to the PSTN, processes the voice and fax signals using
coders/decoders (CODECs) to convert between circuit-switched and packet formats, and works
with the gatekeeper through the Registration Admission Status (RAS) protocol to route calls
through the network. Cisco has numerous VoIP gateway platforms.
With a voice/fax feature card and a T1 or E1 interface card, it can be deployed as the gateway in a
local point of presence (POP).
2. H.323 Gatekeeper
The VoIP gatekeeper in the network is the Multimedia Conference Manager, an H.323-compliant
program implemented as part of the Cisco IOS ® software. The Multimedia Conference Manager
software can run on Cisco 2500 and 3600 routers. The use of gatekeepers makes the network more
scalable by centralizing routing and numbering plan information to facilitate growth and changes.
The gatekeeper resolves addresses, finding the IP address for the gateway configured for that call
destination. It also manages bandwidth and quality-of-service (QoS) requirements. Each
gatekeeper has a “zone” of administrative control, which can control multiple gateways. Such
zones are normally set up to correspond to geographic zones.
3.. RADIUS Server
The RADIUS server performs functions necessary for authentication, authorization, and
accounting (AAA). An ISP can use its existing RADIUS servers the CiscoSecure
implementation for these functions.
32
4. Billing Servers
Billing servers are needed to provide usage-based billing. The ISP may use any RADIUS-based
billing system that supports Cisco VoIP extensions. Examples include Portal, Belle, and Xaact.
The RADIUS server collects and stores call detail records from the VoIP gateways. The billing
servers collect this information from the RADIUS servers and process the data using specialized
billing applications. The final billing statements can be made available to subscribers via the Web
or through the mail, depending on the service provider’s service model.
Local call VoiP Call Local Call
FIGURE 4.: Steps involved in making a long distance VoIP call
���)�� ��������� How the Service Works—Call Processing A typical VoIP long distance call is processed as follows using H.323 protocols:
1. A caller dials the local access number for the gateway. This call arrives at the gateway over an
ISDN or channel associated signaling (CAS) interface from the PSTN.
2. The gateway answers the call.
3. The gateway queries the RADIUS server with the automatic number identification (ANI) of the
caller. The ANI is the caller’s phone number.
�����
���� �����
���� QoS Packet WAN
33
4. The RADIUS server uses the ANI to verify whether the caller is a paying customer. If the ANI
is not in the account database, the gateway prompts for an account or password and sends it to the
RADIUS server for verification.
5. After the RADIUS server has verified that the caller has an account, the gateway plays a second
dial tone for the caller.
6. The gateway collects the destination number digits entered by the caller.
7. The gateway attempts to deliver the call using H.323 by consulting a gatekeeper.
8. The gatekeeper matches the destination number against a set of remote zone E.164 prefixes
configured for the gatekeeper. The match determines the destination zone for the call, which
identifies the far-end gatekeeper.
9. The originating gatekeeper consults with the terminating gatekeeper to select an appropriate
gateway in the remote zone to deliver the call. The address of the remote gateway is then passed
back to the originating gateway using the RAS protocol.
10. The originating gateway places an H.323 call across the IP network to the destination gateway.
If the call attempt fails, the originating gateway tries a different terminating gateway using the
rotary call pattern feature. The originating gateway can be configured to try any number of
alternate routes.
11. The destination gateway places a call using the local PSTN to the call destination.
12. Start/stop records for the call are generated by the originating and terminating gateways and
sent to the RADIUS server.
Facilities embedded in the network:-
User Authentication
Access to the VoIP network can be controlled at the originating gateway using RADIUS-based
authentication. When a caller dials the local access number for the gateway, the AAA interface on
the gateway collects user account information and sends an authentication request to the RADIUS
server.
Interactive Voice Response
The gateway has an integrated interactive voice response (IVR) application that provides voice
prompts and digit collection in order to authenticate the user and identify the call destination.
34
RADIUS Accounting
The VoIP long distance service uses RADIUS-based accounting. The gateway generates start/stop
accounting records for each call leg, which are sent to the RADIUS server to support billing. For a
VoIP call, there are a total of four call legs—an incoming and outgoing call leg at both the
originating and terminating gateways. These legs are linked by a unique 128-bit connection ID.
Each record contains the following information stored in standard RADIUS attributes:
• Calling station ID
• Called station ID
• Call duration
• Received bytes
• Transmitted bytes
• Received packets
• Transmitted packets
4. Call Routing
The gatekeeper routes calls using E.164 prefixes, which are in the form of zone prefixes or
technology prefixes.
35
Regulation: Legal and Policy Issues
$� �� ���� In U.S. telephony, service is split between Local Exchange Carriers (LECs) and InterExchange
Carriers (IXCs). The generally monopolistic LECs provide local telephone service, whereas the
IXCs provide long-distance service between LECs, making up a highly competitive, albeit
regulated, industry.
Most long-distance phone calls in the U.S. involve an LEC connection on both ends (with the
long-distance carrier as the bridge). Each time an IXC terminates or originates a call through an
LEC, the IXC pays the LEC an access charge of roughly 3 cents per minute on each end. This
access charge is greatly inflated but it covers Universal Service obligations. However, in the early
1980's the FCC ruled that providers of 'enhanced services', like Internet Service Providers (ISPs),
need not pay these access charges. ISPs are treated as "end users" who can purchase lines that have
no per minute charge for receiving calls from their customers.
The following is a representative study of a landmark regulatory ruling in the field of VoiP.
America’s Carriers Telecommunication Association (ACTA)5 filed a petition with U.S. Federal
Communications Commission (FCC) to prevent companies from selling Internet Telephony
software and to "institute rulemaking proceedings defining permissible communications over the
Internet".
ACTA argued that it is not public interest to permit long distance services through Internet
Telephony thus depriving those who maintains infrastructure for the same and also it is not in
public interest for these selected communication operators to operate outside regulatory
requirements that are applicable to all other telecommunication provider.
5 ACTA's membership consists primarily of small to medium-sized resellers of long-distance services; larger companies like AT&T, MCI and Sprint are not concerned with ACTA or their petition since they are 'wholesalers of capacity'. Internet telephony is not a form of competition in their market. ACTA's main corporate purpose is to represent these small resellers of long-distance services in legal and political spheres
36
#�� ������� ACTA argued that FCC has regulatory control over the Internet and should take action with regard
to technology of long distance calling. ACTA further argued that goals such as Universal Service
and fair competition in the telecommunications market were being thwarted by Internet
Telephony.
#�� ������� ��Though the petition names VocalTec, Webphone and others as respondents, it was clear that the
FCC's jurisdiction does not extend to software. Netscape, Voxware and Insoft, in a joint opposition
to the petition argued convincingly that in traditional telephony there are many companies who
supply software for operating telecommunications networks - such as software for switching and
signalling in the public switched telephone network – without being subject to FFC regulation. It
would then seem arbitrary to treat Internet Telephony software producers as telecommunications
carriers and other companies who manufacture software for long distance services as not.
Netscape, Voxware and Insoft further suggest that Internet telephone software be, if anything,
customer premises equipment (CPE) since it enable a user's computer and peripheral devices to
communicate over the Internet. Worse still for ACTA (and others), CPE providers are unregulated
and detariffed; state regulation of them has been pre-empted by the Commission itself.
ACTA submitted that Internet Telephony services should be considered interstate
telecommunications carriers under definitions provided in the Telecommunications Act of 1996
#�� 0�� �The U.S. Telecommunications Act of 1996 makes it clear that it is the policy of the United States
Government "to preserve the vibrant and competitive free market that presently exists for the
Internet and other interactive computer services, unfettered by Federal or State regulation". In May
1997, while not explicitly ruling on the ACTA petition, the FCC ruled against requiring ISPs to
pay per-minute access charges - instead an increase in fixed charges on each phone line for
business users was implemented, ISPs included.
37
1��/� 0����� �� �� �����The ACTA petition was fundamentally flawed since it did not identify Internet Telephony as the
actual service (i.e. companies which provide IP Telephony services). It merely identified producers
of software for the output and input of audio, some of whom may have coincidentally been
offering IP Telephony services. The initial definition of Internet Telephony (as merely the use of
the Internet to transmit 'real-time' audio either from PC to PC or from PC to phone) neglects a
third, next generation type of Internet Telephony - phone to phone.
While the first two types of Internet Telephony are inherently tied to the PC (including necessary
software) and Internet Service Providers. The third type, however, is not. In phone-to-phone
Internet Telephony the customer, using an ordinary telephone, dials an access code and then the
telephone number; the access code then routes the call to a special computer gateway (the IP
network). The trouble is that local computer gateways for companies offering this type of service
must be placed in strategic geographic areas. For instance, if a customer using phone-to-phone
Internet Telephony plans to call London (England) from Toronto (Canada), then local gateways
must be located in both London and Toronto. The gateways convert audio into data for
transmission across the IP network and then convert incoming data back into analog signals.
The FCC's definition of phone-to-phone IP Telephony requires that such services:
1.Hold themselves out as providing voice telephony or facsimile transmission service;
2.Do not require the customer to use CPE different from that CPE necessary to place an ordinary
touch-tone call (or facsimile transmission) over the public switched telephone network;
3.Allow the customer to call telephone numbers assigned in accordance with the North American
Numbering Plan (and associated international agreements); and
4. Transmit customer information without net change in form or content.
In the FCC's report to Congress it states that "when an IP telephone provider deploys a gateway
within the network to enable phone-to-phone service, it creates a virtual transmission path between
points on the public switched telephone network. From a functional standpoint, users of these
services obtain only voice transmission, rather than information services such as access to stored
files. Routing and protocol conversion within the network does not change this conclusion,
38
because from the user's standpoint there is no net change in form or content". Given this, together
with the Telecommunications Act's (1996) definitions of a telecommunications carrier,
telecommunications service and 'telecommunications' - 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 - it seems readily apparent that phone-to-phone IP
Telephony companies should be required to pay access charges for connecting to and/or the usage
of the local phone companies' systems. In the absence of a more detailed case-by-case
investigation, however, the FCC withheld any definitive conclusion regarding whether phone-to-
phone IP Telephony should be properly considered a 'telecommunications' rather than an
'information' service.
With regards to specifically PC-to-PC Telephony, the FCC held that "Internet service providers
over whose networks the information passes may not even be aware that particular customers are
using IP telephony software, because IP packets carrying voice communications are
indistinguishable from other types of packets (in which case the) Internet service provider does not
appear to be 'providing' telecommunications to its subscribers." While it would only be fair to
presuppose this will also apply to PC-to-phone IP Telephony, big business cannot make that
assumption.
Stance of European Commission
The European Commission, in supplementing their 1995 Communication on the status and
implementation of the Commission liberalisation Directives, issued a notice on January 15, 1998
(under article 1 of Directive 90/388/EEC) defining its policy on voice telephony in respect of
telephony via the Internet. The notice defines "'voice telephony' means the commercial provision
for the public of the direct transport and switching of speech in real-time between public switched
network termination points, enabling any user to use equipment connected to such a network
termination point in order to communicate with another termination point."6
6 “Status of Voice Communications on Internet under Community Law And, in Particular, Under Directive 90/388/EEC”, Published in the OJ No C 6, 10.1.1998, p. 4. http://europa.eu.int/comm/dg04/lawliber/en/voice_en.htm.
39
Does Internet Telephony falls under this definition of voice telephony?
The European Commission argues that Internet Telephony understood as either PC-to-PC or PC-
to-Phone - is not the principal aim of Internet access providers. The purpose of Internet access is
for the facilitation of browsing, the exchange of electronic mail, and the exchange of data files.
They do, however, properly identify phone-to-phone IP Telephony as involving a commercial
offer. Similar considerations come into play when examining whether Internet Telephony is
correctly "for the public", since computers and access to the Internet are not currently available to
all citizens (nor are there any policies in place, as Universal Service, to help achieve it). The
Commission argues that PC-to-PC Internet Telephony is not available 'for the public', while PC-to-
phone and phone-to-phone IP Telephony are.
The Commission argues that "the time period required (in Internet Telephony) for processing and
transmission from one termination point to the other is generally still such that it cannot be
considered as of the same quality as a standard real-time service" On an IP network packets are
switched; in regular telephony the circuits are. Internet Telephony thus fulfils voice telephony's
stipulation that 'switching' be involved. Internet Telephony is real-time; it simply depends on how
rigorously you define 'real-time'. Surely a delay of no more than two seconds would disbar the
entire technology from being labelled 'real-time'? Even shouting to someone across the street
produces greater delays! (The analogy of cellular communications is also applicable in this
instance).
The Commission concluded that Internet Telephony cannot be properly considered 'voice
telephony' and therefore already falls within the liberalised area, before the deadlines set for the
implementation of full competition. The commission states that with growing sophistication,
certain Internet telephony providers would qualify as providers of voice telephony, and therefore
be subject to the regulatory regime applicable to voice telephony in the future. The Commission
has announced that it will review this policy in light of the evolution of IP Telephony early in
2000. It will be interesting to see if long-distance providers switch to IP networks to avoid
Universal Service contributions.
40
Stance of other Government/Regulators The Mexican government has indicated unofficially that it would apply traditional telephone
regulations and restrictions on phone-to-phone IP voice services, Jacobs continues, but it has not
taken any official action either. A few governments, he says, such as that of Pakistan, have
prohibited personal computer (PC)-based IP voice services, but it doesn't appear to be enforcing
that ban.7
In Singapore, the IP telephone is considered as both the basic telecommunications service and
Internet application, and different regulation methods for different definitions are adopted.The
Internet telephone is completely banned in countries such as Iceland and India. While in Czech and
Hungary, It is specified that the Internet provider must be awarded a license for IP telephony
operations.8
Meanwhile most countries hold wait-and-see attitude, to be decided according to development of
the Internet telephone. Some like Canada have started informally to discuss regulation
methodology for the Internet telephone.
There are two schools of thought regarding IP telephony’s international regulatory future. Many
governments may not worry about it for a while because IP telephony makes up only a small
portion of telecom revenues, and they think it very well may stay that way. Some countries,
however, may be concerned because traditional carriers are using their long distance revenues to
invest in telephone network buildouts. When IP telephony carriers undercut those fees with their 3
cents-a-minute service, the regulators may conclude, it will affect telecom network investments
adversely.
Interoperability of IP Telephone
While the regulators in various countries and regions grapple with the IP Telephony
regulation/deregulation, the industry is striving for interoperability. Interoperability includes two
aspects of contents: interoperability of gateway products and application interoperability. From
the view of current status of interoperability in gateway products, gateway companies now have
products conforming to the H.323, interoperation testing is being performed, and the
interoperability between gateway products produced by many companies will be possible very
7 Kim Sunderland, “ The 1999 Regulatory Outlook for IP Telephony”, www.soundingboard.com. 8 Zang Rong, “Current Status and Future of IP Telephone”, available at www.telecomn.com.
41
soon. To solve the problem of interoperability in gateway products, ITXC, Lucent and VocalTec
have commenced and proposed an interoperability frame of i.NOW!
Thus it appears that in the absence of sound regulatory (or deregulatory stance) by the various
regulators , industry itself is moving towards self-regulation and proposing compatible standard.
In other words, we strongly feel that technology would drive regulation, by means of evolving
technical solutions for problems even when semantical debates are being carried out in courts of
law.
42
INTERNET TELEPHONY IN INDIA
The sub-committee on telecommunications, headed by Planning Commission advisor Montek
Singh Ahluwalia, is likely to recommend liberalisation of Internet telephony or voice on Internet
(VoI) norms, paving the way for issuing Internet telephony licenses in a year. In the New Telecom
Policy 1999, the government had not permitted Internet telephony. However, the decision was kept
subject to review. Similarly, in its guidelines for Internet service providers (ISP), the department of
telecommunications (DoT) had said that the licence would be liable for termination following any
violation. Experts believe that there may be some opposition from the Videsh Sanchar Nigam Ltd,
as Internet telephony will end its monopoly. Also, Communications Minister Ram Vilas Paswan
has said9 that Net telephony would strictly be kept off the reach of private ISPs to safeguard
government revenues.
A study by a London-based telecom consultancy had projected in 1997 that VSNL would
lose $54 million (about Rs 195 crore) by the year 2001 if Net telephony were legalized. VSNL has
until now blocked out sites on the Net which offer voice telephony services.
Small players in India (e.g., Premiere Infosystems, a Noida tech start-up) have started offering
installation of software for voice services over the Net at less than 50 paise per minute. To put this
in perspective: an international one-minute call on a normal basic telephone would cost some Rs
75. The infrastructure required at the subscribers’ end is remarkably limited: a computer-
multimedia kit and Net telephony software. Apart from a computer and a high-speed modem, users
indulging in Netspeak need to install a sound-card (which converts sound into electronic signals
and vice versa), a microphone, speakers and Net telephony software. The software like Net-To-
Phone can be easily downloaded from the Net.
The cost of installing the software is less than Rs 2,000, after which every call domestic or
international long distance would be billed as a local call: maximum of Rs 1.40 per minute. Calls
could be made to other telephones or to other computers rigged up with a multimedia kit.
9 Source: Cyber News Service , 7/19/00.
43
Significant interest has been shown by various IP telephony players in India, despite the
regulations. In July’00, Israel based Arelnet Limited, leading provider of IP telephony solutions,
expressed its intentions of launching VoIP gateways in India, for small and medium-sized ISPs.10
,(��� (1 �0(!'1�0�#'(� The modes of proliferation of VoIP in India are primarily three
1. Traditional Dial Up Internet (through an ISP)
2. Cable access (through a cable modem through existing cable TV lines)
3. Phone to Phone (Major providers could be existing established players like VSNL and
DOT)
1.Traditional Dial Up Internet
Internet Service Providers connect customers to the Internet. For a particular access fee, the
service provider provides an installation software, a username and password and access phone
number.
National ISPs
These ISPs operate points of presence throughout the country. One category of national
ISPs own the network backbone and lease the international connectivity while the other category
of players lease the network and the international connectivity from other ISPs.
The main target segment for these ISPs is the corporate segment. A presence throughout
the country helps these ISPs to cover all the locations for a particular corporate. ISPs owning the
network backbone use the reliability of service (due to ownership of the network) as a key
USP to customers. This is helpful, especially for real-time mission-critical applications.
Regional and local ISPs
These are the ISPs that either operate in the smaller towns or particular states. These ISPs
serve both the business and consumer segments usually within a geographic region.
THE ISP SCENARIO IN INDIA Internet subscribers in India are expected to reach a figure 5,30,000 by March this year,
according to IDC. This figure is expected to touch 1.3mn by March 2001, a growth of 145% over
the previous year. Dataquest (DQ) estimates are slightly more optimistic. DQ expects the Internet
subscriber base to reach 6,55,000 by March 2000. DQ expects this to shoot up to 1.86mn by 2001
10 Israel IP firm enters India, Business Standard, July 30th, 2000.
44
and 3.75mn by 2002. DQ’s estimates say that the ISP access market was worth Rs1.02bn in 1999
and is likely to touch Rs2.13bn in 2000. The market will expand sharply to touch Rs8.37bn in
2001 and to Rs15bn by 2002.
Falling PC prices, coupled with a drop in the access rates (due to the price war in the ISP
market) has caused this growth. To some extent, future growth will also be assisted by Internet
access through cable TV. Estimates are that three to four years down the line, India, with 30mn
Internet users, will have the most Internet users in Asia, next only to China.
The last 12 months has seen a good growth in the usage of Internet by the corporate
segment also. The factors, which have helped this, is the lowering of leased line charges by TRAI
and the offering of a variety of value-added services by the ISPs. While earlier the corporates were
using the Internet more as an information provider – email, surfing etc, many of them are using
this to do some kind of e-commerce especially on the vendor end.
According to the IDC survey on 31st July 1999, the small / medium organizations had the
largest portion of the subscriber base. Large organizations were next, with a subscriber base of
nearly 97,000. The home segment showed a healthy growth in its subscriber base share over its
November 1998 share of 8.9%. In the next few years, the home and small/ medium sized segments
will experience unprecedented growth in terms of Internet connectivity.
� PC / Internet Penetration comparison
India US Asia-Pacific
Population 1000 270.3 2769.6
Number of Internet users 0.5 62.8 10.2
Net-enabled PCs 0.3 87.4 9.5
Internet users / population 0.1% 23.2% 0.4%
Net-enabled PC / household (%)11 0% 129% 1%
Growth of the corporate access market is also expected with large domestic
computerization measures in the Government sector. Already initiatives are underway on the part
of the Government to start a comprehensive move towards electronic governance. The corporate
11 assuming family size of 5 in India and 4 otherwise; IDC, Indiainfoline estimates
45
access market will also grow as the economy gathers momentum and companies realise the
benefits of using the Internet.
2.Cable ISPs
These ISPs are normally owned by the cable companies that help them to get exclusive use
of the cables. (The cable IP telephony technology has been explained in detail earlier)
Cable access ISPs offer broadband Internet connectivity through the coaxial fiber networks. Cable
ISPs usually service the consumer segment of the market. The parent makes money from the lease
charges of the cable while the ISP uses the access / e-commerce / advertising revenues to
compensate for the hiring charges.
The development of alternative means of access is also expected to give Internet
usage a boost. The most obvious way is using the TV Cable network, given the enormous TV
penetration in India. Currently, 30mn Indian households have a TV. If this segment of the
population can be tapped, Internet usage can explode. Currently, set-top boxes, which connect
users to the Internet through the TV, are quite costly. However, with increasing penetration the
cost of these boxes should fall and help increase ISP demand.
Indian Cable Television Industry
. The cable industry however is growing at what some consider a chaotic rate; because
entrance barriers are low, the industry is open to almost anyone. Currently there exist over 100,000
cable operators in India (compared to 10,000 in the US) employing over 1 million people. Ninety-
seven percent of cable operators have less than 1,000 subscribers, most having less than 500.
Presently, an initial investment of Rs.250,000 ($7,287) covering a dish, signal receiver, mixer,
amplifiers and a modulator to convert frequency is all that is needed. The only other expenses are
installation and small VCR movies fee: all satellite signals are received free. Eighty percent of
Indian cable systems carry between six and eight channels, fifteen percent between ten and twelve,
and five percent carry more than twelve. Over 30 million Indian households receive cable serving
an audience of well over 125 million, the second largest cable market in Asia after China. For this
reason, the Cable Television Networks (Regulation) Bill was presented to the Indian Parliament
and passed December 13, 1994.
Hence, the medium will be a powerful mode of Internet penetration, especially of
broadband services. Internet over cable is picking up very fast in India this year. Some of the
major service providers are Hathaway in Mumbai and Spectranet in Delhi. However as compared
46
to the traditional dial-up services, the subscriber charges are around Rs.1000 per month for
unlimited access. These are in addition to the cost of a cable modem.
Once IP telephony is permitted in India, we feel that Cable IP will be a major mode of
proliferation of IP telephony services. The current subscription rates of Internet over cable are
targeted towards corporates and high-income consumers. However, subscribers may retail these
services to other customers (who can’t purchase these services themselves) in future “IP booths”.
Also, these services may be combined with “ cyber cafes” which already provide Internet services
to customers.
Lack of a proper telecom infrastructure has kept the Internet penetration down to only
the major towns and cities. With the setting up of a National Telecom Backbone, it is expected that
Internet will reach more areas of the country. Higher bandwidth availability is also expected to
spur the usage of Internet in terms of both number of subscribers as well as the usage time by a
particular user.
Key Features of the Indian Internet Policy
1. Licensee has the freedom to lease domestic backbone from DOT, basic service providers,
SEBs, the power grid corporation, railways or any other authorized operator.
2. License fee is absent for the first five years and is nominal (Re.1) after that.
3. The decision on tariffs has been left to the ISP providers. However, TRAI has the right to
review and fix the tariff anytime during the licensee period of a licensee.
4. The ISP policy does not allow for Internet telephony.
The government has issued in principle clearance to all ISPs who have applied for
international gateway license using satellite technology. The government has given around 225
ISP licenses so far.
47
2�� ���� ��� ������ ' ��� '� #����� � ������ ��� ���+
Retail Segment In the retail market, the competition is likely to be fierce because the only service
provided is a plain vanilla service and, therefore, there is very little differentiation, which is
possible. So the competition would be around price.
Since extent of value addition is very low here, the only way to compete is by reaching
economic volumes, so it is essential to make large upfront investments in infrastructure and service
and capture as much of the market as possible.
Corporate Segment Corporate customers have relatively lower price elasticity towards access charges. As most of
them use the voice and facsimile communication to support business critical applications their
focus is on service rather than costs. Corporate customers are easy to retain because service
disruptions due to a change in vendor can be extremely costly. Thus, entry barriers in this market
are very high. This is also because a good track record is essential to get business from corporates.
With e-commerce becoming a buzzword, IP telephony helps in creating a completely new class of
service such as web-enabled call centers, telecommuting and long distance learning
����� � !��� � � ���� �� �� � ' ' ��� Presently IP Telephony is banned in India. Unlike US there has not been any comprehensive
debate on the issue. Following are some of the factors that are driving the government stance:
1. International Accounting Rate System
As per the prevalent International Accounting Rate System, India benefits from heavy incoming
calls compared to outgoing calls. Under this system whichever country has net incoming calls gets
paid for the same. The United States sends billions of dollars abroad as a result of such
international settlement rates and a significant chunk of it comes to India. While IP telephony
could save America billions of dollars, possibly a significant portion of the size of a federal
universal service fund, India stands to lose out the precious foreign exchange, which it presently
gets due to exorbitant rates being charged by VSNL from consumers making international calls.
2. Effect over investment in traditional telecom
48
After liberalisation of telecom sector in 1994, heavy investment has been made in private and
public sector. These operators of basic and cellular telephony are required to pay heavy license
fees in the tune of thousands of crores every year. If IP Telephony is allowed to become popular
and unregulated, it will have adverse impact on the present telecom player and not only the
investment made in developing telecom structure may parish but even government may lose out a
major source of revenue.
3. Level Playing Field
Another major concern of Indian Regulators is to provide level playing field to traditional
telephony operators and IP telephony operators.
3��� ��� ����� ��� ��� ' ��.� ��+ A complete scenario of Indian telephony is not possible without considering the options
before the incumbents, MTNL, DTS and DOT when voice services over the Internet are allowed.
For this purpose, the strategy adopted by the largest U.S. telecommunication provider, AT&T in
the wake of IP Telephony was studied.
The Telecommunications Act of 1996, which allowed local and long distance phone companies,
Internet firms, and cable businesses to compete in each others’ markets, immediately led way to a
long list of mergers and acquisitions.
AT & T has set itself the objective of becoming the leader in end-to-end communication
services. In March 1999, AT&T made a $62 billion cash bid for MediaOne, one of the largest
suppliers of broadband services on cable. The driving force for this merger is that AT&T wants to
take advantage of recent consolidation in the industry and changing technology. The purchase of
the Cable Company was not only intended to bypass the Regional Bell Operating Companies12, but
also to deliver a full menu of phone, Internet and multimedia entertainment.
If the deal goes through, it will create the largest Cable Company in the U.S., with over 16
million subscribers. It is now planning to provide Cable IP telephony services.
Hence, it is clear from the above moves that AT&T is unfazed about the arrival of new
technologies as substitutes for its services. Instead it has proactively adopted these new
technologies so that it is not left behind. It plans the migration of its services from the old
technology to the new as it upgrades its network across the U.S. for digital services.
12 Local access providers in the U.S.
49
We believe that it is futile for the incumbents in India to restrain the growth of IP telephony
as it may not be possible to restrain users from accessing and using the services in light of the
significant price differences. Given the fact that international gateways for private ISPs have been
allowed, it would become more difficult to regulate the rendering of these services. In contrast, the
incumbents should use their existing infrastructure and first-mover advantage to enable themselves
to expedite these services on their systems. Such a process would involve the following: -
� Technology assessment
� Establishment of Additional Infrastructure (Gateways, Gatekeepers, etc.)
� Deployment of Software
50
PRICING OF IP TELEPHONY The issue of pricing IP Telephony is closely linked with the pricing of Internet services as a whole
if viewed as capable of providing different kinds of services to the users.
One opinion is that we have survived so far. There is rather substantial experience with the current
model, and it seems to meet the needs of many users. However, others believe that once
commercial Internet service becomes mature, customers will start to have more sophisticated
expectations, and will be willing (and indeed demand) to be able to pay differential rates for
different services. Indeed, there is already evidence in the marketplace that there is a real need for
service discrimination. The most significant complaint of real users today is that large data
transfers take too long, and that there is no way to adjust or correct for this situation. People who
would pay more for a better service cannot do so, because the Internet contains no mechanism to
enhance their service.
In case of IP Telephony this debate becomes more critical because “voice packets” generated from
IP Telephony users needs to be routed fast the other end to maintain semblance of real time
communication. For this priority routing the “voice packets” needs to be given priority by the
routing mechanisms.
In 1994 Jeffrey K. MacKie-Mason and Hal R. Varian (University of Michigan) advocated the
usage based pricing of Internet and presented the concept of “smart market”. They proposed a way
to price network usage that they called ``smart markets.'' Much of the time the network is
uncongested; at such times the price for usage should be zero. However, when the network is
congested, packets are queued, delayed, and dropped. The current queuing scheme is FIFO. They
propose instead that packets should be prioritized based on the value that the user puts on getting
the packet through quickly. To do this, each user assigns her packets a bid measuring her
willingness-to-pay for immediate servicing. At congested routers, packets are prioritized based on
bids. In order to make the scheme incentive-compatible, users are not charged the price they bid,
but rather are charged the bid of the highest priority packet that is not admitted to the network.
(Vickrey Auction) It can be shown that this mechanism provides the right incentives for users to
reveal their true priority.
51
David D. Clark (1995)13 also discusses a scheme where bandwidth is allocated among users by
creating different service classes of different priorities to serve users with different needs. The
definition of priority is that if packets of different priority arrive at a switch at the same time, the
higher priority packets always depart first. This has the effect of shifting delay from the higher
priority packets to the lower priority packets under congestion. If this argument is pushed further
by redefining priority where high priority packets (voice packets) are sent first even if they arrive a
little later than the other packets (normal data packets).
Clark further says that the slowing down an individual packet does not much change the observed
behavior. But the probable effect of priority queuing is to build up a queue of lower priority
packets, which will cause packets in this class to be preferentially dropped due to queue overflow.
The rate adaptation of TCP translates these losses into a reduction in sending rate for these flows
of packets. Thus, depending on how queues are maintained, a priority scheme can translate into
lower achieve throughput for lower priority classes.
This might, in fact, be a useful building block for explicit service discrimination except that
priority has no means to balance the demands of the various classes. The highest priority can
preempt all the available capacity and drive all lower priorities to no usage. In fact, this can easily
happen in practice. A well tuned TCP on a high performance workstation today can send at a rate
exceeding a 45 mb/s DS-3 link. Giving such a workstation access to a high priority service class
could consume the entire capacity of the current Internet backbone for the duration of its transfer.
It is not likely that either the service provider or the user (if he is billed for this usage) desired this
behavior.
The other drawback to a priority scheduler (where priority is not based on price) for allocating
resources is that it does not give the user a direct way to express a desired network behavior. There
is no obvious way to relate a particular priority with a particular achieved service. Most proposals
suggest that the user will adjust the requested priority until the desired service is obtained. Thus,
the priority is a form of price bid and not a specification of service (which is similar to the smart
13 David D. Clark, “A Model for Cost Allocation and Pricing in the Internet”, Presented at MIT Workshop on Internet Economics March 1995.
52
market pricing proposed by Hal R. Varian (1994)). It is much more direct way to let the user
directly specify the service he desires, and let the network respond.
Above concept of usage based pricing may be the fundamental of IP Telephony pricing where 1 IP
Telephony voice packet transmission (with priority routing) price will be equal to x normal data
packet transmission price. At times when network is uncongested and spare capacity is available
the value of x will be around 1 but when the network is congested and capacity is being utilized
fully, the value of x will high and cost of may even approach cost of long distance telephony on
traditional network.
However, such dynamic pricing and billing for every packet may not be possible for IP Telephony
service provider (ISP, Cable operator) but this can be approximated by putting different slabs for
different values of x based on historical congestion (magnitude and timings). This mechanism is
somewhat similar to what is being followed for peak and off-peak hours in traditional telephony.
���� �� '� #����� � � ' ��� Though it is difficult to develop the above-mentioned dynamic model, static models where values
of x can be varied depending over network congestion can approximate it.
In this section the cost of IP telephony if implemented in India has been estimated based on the
following explicit assumptions:
Key Assumptions � 64 kbps ISDN connection for one IPT connection (this is very estimates. In fact 64 kbps should
be sufficient to allow user to make two IP telephony calls at the same time. This conservative
estimate has been taken in view of congested network condition in India that prevails most of
the time and it is felt that 64 kbps connection will make the last mile network congestion free
for good quality voice communication by providing excess bandwidth at user end. Another
reason for taking ISDN costs is that capital investment very closely approximates capital cost
of IP Telephony ports for phone to phone telephony)14.
14 Partly based on methodology followed by Andrew Sears (1996) for US market where he used two line of 14.4 kbps each for one IP Telephony lines. However, technical feasibility of combining two low capacity line and developing high capacity line is yet to be proven sound. Improved for additional requirement of IP Telephony. See Andrew Sears, “The Effect of Internet Telephony on the Long Distance Voice Market”, 1996.
53
� 1 data packet of IPT is equivalent to x nos. of normal data packet (as discussed earlier this
variable is a function of network congestion, which may vary over time in a day).
� Fixed cost recovery in 4 years (Again it appears a bit conservative but has been taken as the
rate of change of technology is quite fast and may make investment obsolete).
� Cost estimates are for ISP.
� Local call charges Rs. 1 per minutes (to make it unsubsidized).
� 1000 data hours have been assumed for a year and price has been taken for the same per ISDN
connection.
0����� Yearly Cost of ISDN connection (1000 hours) = 38033 Rs. (See Exhibit 1 for detailed
calculations)15.
Priority Premium
Factor
Equivalence (in Rs.)
X = 1 1 DH = 1 IPTH 1 hour cost of IPT 38.03333
X = 2 2 DH =1 IPTH 1 hour cost IPT 76.06667
X = 3 5 DH = 1 IPTH 1 hour cost IPT 190.1667
DH = Data Hour IPTH = IP Telephony Hour Cost of IP Telephony for various values of x: (all figures are in Rs.)
x Without Local Call With Local Call
Voice
Price
Voice
Price
Voice
Price
Voice
Price
(1 IPT packet eq.
Of x data packet
(per hour) (per min) (per hour) (per min)
1 38.03 0.63 98.03 1.63
2 76.07 1.27 136.07 2.27
5 190.17 3.17 250.17 4.17
15 Costs for estimation have been taken from VSNL and MTNL websites. The rates are for basic access from MTNL for ISDN connection and data hours rates are from VSNL.
54
From the above analysis it is evident that the IP telephony is going to be significantly cheaper that
traditional telephony (see VSNL rates for overseas calls and ISD rates in India in Exhibit 2).
How cheap is IP Telephony compared to traditional telephony?
To find out the actual discount available from IP telephony at present, prices of per minute call
from US to 180 countries for IP telephony were compared with the traditional telephony prices
(Exhibit 3). Based on analyses of pricing data (US to180 countries) available from PC to phone
companies (net2pnone and deltathree) with traditional telephony provider (Telecom International
Ltd.), it was found that IPT is around 25% (with stdev. 13%) cheaper on average (for major
destination it is 40-50%). This discount can be increased as presently part of the path is on
traditional telephony network. Once a seamless phone to phone VOIP connection infrastructure is
in place, this price is further expected to come down.
Major Hurdles
One of the major hurdles (other than clear regulatory framework) in roll out of VOIP network is
high cost of IP Telephony port ($ 1000 - 2000 compared to $100-200 for traditional telephony
port). Though it is likely to come down in near future, it is also to be considered that cost of
traditional telecom equipment is also falling.
Possible Answers
One of the positive approaches to make most of existing VOIP network is to use yield
management in IP Telephony.16 It is true that providing 100% available connection to all the users
for 24 hours will not be possible at a cheaper cost and the capital investment will be prohibitive.
This can be overcome using different values of x for different network congestion conditions. This
approach coupled with falling hardware prices is can make IP Telephony a cost effective solution
for the communication for the masses.
16 Brett A. Leida, “A Cost Model Of Internet Service Providers: Implications For Internet Telephony And Yield Management”, MIT, 1998.
55
�4��.�� 56 ���� ��������� ��� � ���� ��� '� #����� �
(All Figures are in Rs.) MTNL Cost (Rs.)
One Time Cost 17100
Yearly One time cost 5700 (cost recovery in 4 years @ 15.5% cost of capital)
Yearly Charges 12000
VSNL Cost
One Time Cost 1000
Yearly One time cost 333.33 (cost recovery in 4 years @ 15.5% cost of capital)
Yearly Charges 20000
(for 1000 data hours)
Yearly Charges (1000 hrs) 38033.33 (excluding local calls)
The above costs are based on rates available on www.vsnl.net.in and www.mtnl.com .
56
�4��.�� 76 ���)��� � �#� � � '�� ����� � ' �����
(from Oct 1, 2000 in Rs./min)
Distance Pulse Charges @ Rs. 0.80 @ Rs. 1.00 @ Rs. 1.20
STD Rates Up to 50 km 0.80 1.00 1.20 50 – 200 km 4.00 5.00 6.00
201 – 500 km 8.00 10.00 12.00 500 – 1,000 km 12.00 15.00 18.00 Above 1,000 km 16.80 21.00 25.20
ISD Rates SAARC & Other Neighboring
Nations 16.80 21.00 25.20
Africa, Europe, Gulf, Asia and Oceania
27.20 34.00 40.80
America & Western hemisphere
32.40 41.00 49.20
17 Anonymous, “ Trai Slashes domestic, overseas telecom charges by 16-23%”, The Economic Times, 29th August, 2000.
57
�4��.�� 86 ��������� �� ����� �� '� #����� � ��� �� ��� �� ���� #������� �
#����� � ����� �$� ����� ��� �)������ ������ ����
Country Net2Phone Deltathree Smartlink Tele. Int. Average for
IPT Average for Traditional Tele.
Discount
Albania $0.27 $0.32 $0.49 $0.30 $0.29 $0.40 26% Algeria $0.24 $0.47 $0.69 $0.39 $0.35 $0.54 35% American Samoa $0.28 $0.31 $0.76 $0.53 $0.30 $0.65 54% Angola $0.45 $0.43 $0.91 $0.57 $0.44 $0.74 41% Argentina $0.39 $0.39 $0.59 $0.25 $0.39 $0.42 7% Armenia $0.59 $0.60 $0.99 $0.35 $0.59 $0.67 11% Aruba $0.23 $0.39 $0.53 $0.35 $0.31 $0.44 30% Ascension Island $0.70 $0.84 $1.09 $0.84 $0.77 $0.97 20% Australia $0.08 $0.16 $0.21 $0.08 $0.12 $0.15 17% Austria $0.10 $0.23 $0.36 $0.08 $0.16 $0.22 25% Azerbaijan $0.44 $0.50 $0.65 $0.49 $0.47 $0.57 18% Bahrain $0.63 $0.61 $0.99 $0.80 $0.62 $0.90 31% Bangladesh $0.69 $0.88 $1.32 $0.75 $0.78 $1.04 24% Belarus $0.39 $0.45 $0.55 $0.30 $0.42 $0.43 1% Belgium $0.08 $0.20 $0.30 $0.08 $0.14 $0.19 27% Belize $0.64 $0.66 $0.99 $0.71 $0.65 $0.85 23% Bhutan $0.47 $0.61 $1.15 $0.33 $0.54 $0.74 27% Bolivia $0.53 $0.60 $0.88 $0.42 $0.56 $0.65 13% Bosnia And Herzogovina
$0.36 $0.43 $0.64 $0.30 $0.40 $0.47 16%
Botswana $0.29 $0.38 $0.79 $0.46 $0.34 $0.63 46% Brazil $0.22 $0.33 $0.59 $0.21 $0.27 $0.40 32% Bulgaria $0.25 $0.45 $0.50 $0.33 $0.35 $0.42 16% Burkina Faso $0.57 $0.68 $0.84 $0.64 $0.62 $0.74 16% Burundi $0.58 $0.54 $0.97 $0.69 $0.56 $0.83 33% Cambodia $0.79 $1.03 $1.35 $1.03 $0.91 $1.19 24% Cameroon $0.65 $0.67 $1.02 $0.85 $0.66 $0.94 30% Canada $0.04 $0.15 $0.19 $0.12 $0.09 $0.15 39% Cape Verde Isl $0.45 $0.61 $0.84 $0.58 $0.53 $0.71 25% Central African $0.85 $0.74 $1.16 $0.99 $0.79 $1.08 26% Chad $0.99 $0.90 $1.40 $1.27 $0.95 $1.34 29% Chile $0.17 $0.25 $0.39 $0.11 $0.21 $0.25 16% China $0.25 $0.43 $0.92 $0.21 $0.34 $0.57 39% Colombia $0.22 $0.32 $0.69 $0.15 $0.27 $0.42 36% Comoros $0.75 $0.66 $1.17 $1.02 $0.71 $1.10 35% Congo $0.69 $0.51 $1.09 $0.75 $0.60 $0.92 35% Cook Islands $0.95 $1.07 $1.50 $1.15 $1.01 $1.33 24%
Costa Rica $0.29 $0.31 $0.69 $0.43 $0.30 $0.56 46% Croatia $0.28 $0.41 $0.54 $0.26 $0.34 $0.40 14%
18 IP Telephony prices have been taken from www.net2phone.com and www.deltathree com while the traditional telephony prices have been taken from www.smartlink,com and www.telecominternationl.com.
58
Cyprus $0.31 $0.34 $0.57 $0.26 $0.32 $0.42 22% Czech Republic $0.22 $0.32 $0.42 $0.17 $0.27 $0.30 8% Denmark $0.08 $0.15 $0.26 $0.08 $0.11 $0.17 33% Djibouti $0.73 $0.90 $1.10 $0.93 $0.82 $1.02 20% Dominican Republic
$0.15 $0.30 $0.39 $0.23 $0.23 $0.31 27%
Ecuador $0.26 $0.63 $0.78 $0.34 $0.45 $0.56 20% Egypt $0.54 $0.72 $0.99 $0.60 $0.63 $0.80 21% El Salvador $0.34 $0.43 $0.69 $0.36 $0.39 $0.53 26% Eritrea $0.99 $1.09 $1.50 $0.99 $1.04 $1.25 16% Estonia $0.25 $0.38 $0.44 $0.20 $0.32 $0.32 1% Ethiopia $0.79 $0.98 $1.30 $0.96 $0.88 $1.13 22% Falkland Islands $0.38 $0.39 $1.07 $0.71 $0.38 $0.89 57% Fiji Islands $0.83 $0.82 $1.18 $0.99 $0.83 $1.09 24% Finland $0.10 $0.21 $0.28 $0.08 $0.15 $0.18 15% France $0.08 $0.19 $0.25 $0.08 $0.13 $0.17 19% French Guiana $0.39 $0.61 $0.65 $0.40 $0.50 $0.53 4% French Polynesia $0.63 $0.58 $0.92 $0.66 $0.60 $0.79 24% Gambia $0.43 $0.57 $0.75 $0.58 $0.50 $0.67 25% Georgia $0.65 $0.55 $0.90 $0.35 $0.60 $0.63 4% Germany $0.08 $0.18 $0.25 $0.08 $0.13 $0.17 23% Ghana $0.31 $0.49 $0.82 $0.49 $0.40 $0.66 39% Gibraltar $0.35 $0.34 $0.54 $0.37 $0.35 $0.46 24% Greece $0.19 $0.33 $0.49 $0.20 $0.26 $0.35 24% Greenland $0.36 $0.70 $0.70 $0.54 $0.53 $0.62 14% Guadeloupe $0.38 $0.46 $0.67 $0.50 $0.42 $0.59 29% Guam $0.10 $0.16 $0.29 $0.12 $0.13 $0.21 38% Guatemala $0.25 $0.41 $0.69 $0.26 $0.33 $0.48 31% Guinea $0.40 $0.48 $0.96 $0.59 $0.44 $0.78 43% Guyana $0.76 $0.99 $1.25 $0.87 $0.87 $1.06 18% Haiti $0.53 $0.68 $0.85 $0.66 $0.60 $0.76 20% Hong Kong $0.08 $0.16 $0.36 $0.08 $0.12 $0.22 46% Hungary $0.19 $0.33 $0.37 $0.18 $0.26 $0.28 6% Iceland $0.13 $0.25 $0.46 $0.26 $0.19 $0.36 47% India $0.49 $0.81 $0.99 $0.55 $0.65 $0.77 16% Indonesia $0.29 $0.40 $0.94 $0.36 $0.35 $0.65 47% Iran $0.80 $0.90 $1.17 $0.95 $0.85 $1.06 20% Iraq $0.89 $1.05 $1.24 $1.11 $0.97 $1.18 18% Ireland $0.08 $0.23 $0.29 $0.08 $0.15 $0.19 16% Israel $0.09 $0.24 $0.49 $0.08 $0.17 $0.29 42% Italy $0.10 $0.23 $0.37 $0.08 $0.17 $0.23 27% Ivory Coast $0.70 $0.80 $1.17 $0.89 $0.75 $1.03 27% Jamaica $0.45 $0.64 $0.81 $0.46 $0.54 $0.64 14% Japan $0.08 $0.21 $0.39 $0.08 $0.14 $0.24 39% Jordan $0.59 $0.76 $0.99 $0.85 $0.67 $0.92 27% Kazakhstan $0.49 $0.65 $0.89 $0.35 $0.57 $0.62 8% Kenya $0.63 $0.71 $1.07 $0.78 $0.67 $0.93 28% Kiribati $0.83 $0.86 $1.19 $0.99 $0.85 $1.09 22% Kuwait $0.59 $0.77 $1.15 $0.49 $0.68 $0.82 17% Kyrgystan $0.54 $0.60 $0.94 $0.35 $0.57 $0.65 12%
59
Laos $0.73 $0.86 $1.42 $0.89 $0.80 $1.16 31% Latvia $0.33 $0.44 $0.50 $0.38 $0.38 $0.44 13% Lebanon $0.58 $0.62 $0.99 $0.68 $0.60 $0.84 28% Lesotho $0.43 $0.48 $0.95 $0.57 $0.46 $0.76 40% Liberia $0.39 $0.46 $0.73 $0.49 $0.43 $0.61 30% Libya $0.36 $0.43 $0.63 $0.39 $0.40 $0.51 22% Luxembourg $0.10 $0.24 $0.30 $0.08 $0.17 $0.19 11% Macao $0.39 $0.40 $0.73 $0.56 $0.39 $0.65 39% Macedonia $0.35 $0.51 $0.57 $0.47 $0.43 $0.52 18% Madagascar $0.79 $0.70 $1.18 $1.00 $0.75 $1.09 32% Malawi $0.45 $0.46 $0.63 $0.48 $0.46 $0.56 18% Malaysia $0.10 $0.29 $0.49 $0.19 $0.20 $0.34 42% Maldives $0.69 $0.73 $1.10 $0.83 $0.71 $0.97 26% Mali $0.79 $0.89 $1.15 $0.89 $0.84 $1.02 18% Malta $0.22 $0.32 $0.48 $0.24 $0.27 $0.36 26% Mauritania $0.58 $0.44 $0.99 $0.72 $0.51 $0.86 40% Mauritius $0.59 $0.89 $0.85 $0.72 $0.74 $0.79 6% Mexico $0.16 $0.29 $0.49 $0.17 $0.23 $0.33 32% Moldova $0.44 $0.70 $0.88 $0.73 $0.57 $0.81 29% Monaco $0.11 $0.28 $0.33 $0.35 $0.20 $0.34 42% Montserrat $0.64 $0.66 $0.80 $0.87 $0.65 $0.84 22% Morocco $0.39 $0.54 $0.64 $0.47 $0.46 $0.56 16% Mozambique $0.48 $0.50 $0.88 $0.72 $0.49 $0.80 39% Namibia $0.38 $0.35 $0.89 $0.56 $0.37 $0.73 49% Nauru $0.70 $0.82 $1.34 $1.05 $0.76 $1.20 37% Nepal $0.78 $0.94 $1.29 $0.95 $0.86 $1.12 23% Netherlands $0.08 $0.22 $0.24 $0.08 $0.15 $0.16 6% New Caledonia $0.64 $0.68 $0.95 $0.75 $0.66 $0.85 23% New Zealand $0.08 $0.17 $0.30 $0.08 $0.12 $0.19 34% Nicaragua $0.44 $0.61 $0.82 $0.56 $0.53 $0.69 24% Niger $0.69 $0.67 $1.11 $0.69 $0.68 $0.90 24% Nigeria $0.69 $0.85 $0.99 $0.74 $0.77 $0.87 11% Niue $0.95 $1.12 $1.49 $1.16 $1.03 $1.33 22% Norway $0.08 $0.18 $0.25 $0.08 $0.13 $0.17 22% Oman $0.85 $0.71 $1.19 $0.90 $0.78 $1.05 26% Pakistan $0.49 $0.90 $1.27 $0.78 $0.69 $1.03 32% Palau $0.70 $0.37 $49.00 $0.89 $0.54 $24.95 98% Panama $0.49 $0.57 $0.85 $0.57 $0.53 $0.71 26% Paraguay $0.59 $0.61 $0.95 $0.41 $0.60 $0.68 11% Peru $0.25 $0.49 $0.89 $0.31 $0.37 $0.60 39% Philippines $0.21 $0.40 $0.75 $0.21 $0.31 $0.48 36% Poland $0.16 $0.36 $0.42 $0.29 $0.26 $0.36 27% Portugal $0.15 $0.32 $0.50 $0.16 $0.24 $0.33 28% Puerto Rico $0.07 $0.16 $0.19 $0.07 $0.11 $0.13 13% Qatar $0.69 $0.69 $1.10 $0.79 $0.69 $0.95 27% Reunion Island $0.55 $0.45 $0.97 $0.69 $0.50 $0.83 40% Romania $0.39 $0.50 $0.59 $0.35 $0.44 $0.47 5% Rwanda $0.79 $0.82 $1.19 $0.83 $0.81 $1.01 20% San Marino $0.19 $0.78 $0.57 $0.45 $0.49 $0.51 5%
60
Singapore $0.11 $0.26 $0.42 $0.15 $0.19 $0.29 35% Slovakia $0.25 $0.38 $0.42 $0.24 $0.31 $0.33 5% Slovenia $0.24 $0.38 $0.44 $0.19 $0.31 $0.32 2% South Africa $0.29 $0.43 $0.69 $0.30 $0.36 $0.50 27% Spain $0.10 $0.34 $0.47 $0.08 $0.22 $0.28 20% Sri Lanka $0.86 $0.79 $1.27 $0.81 $0.83 $1.04 20% St. Helena $0.65 $0.99 $0.97 $0.81 $0.82 $0.89 8% Sudan $0.43 $0.52 $0.71 $0.52 $0.47 $0.62 23% Suriname $0.98 $0.67 $1.35 $1.11 $0.83 $1.23 33% Swaziland $0.23 $0.33 $0.60 $0.27 $0.28 $0.44 36% Sweden $0.08 $0.16 $0.20 $0.08 $0.12 $0.14 14% Switzerland $0.10 $0.19 $0.27 $0.08 $0.15 $0.18 17% Syria $0.67 $0.72 $1.15 $0.84 $0.69 $1.00 30% Taiwan $0.10 $0.24 $0.63 $0.11 $0.17 $0.37 55% Tajikistan $0.47 $0.53 $0.89 $0.35 $0.50 $0.62 19% Tanzania $0.49 $0.58 $1.07 $0.62 $0.54 $0.85 37% Thailand $0.49 $0.46 $0.89 $0.45 $0.47 $0.67 29% Togo $0.85 $0.98 $1.19 $0.89 $0.92 $1.04 12% Tonga Islands $0.96 $1.01 $1.29 $0.98 $0.98 $1.14 13% Tunisia $0.35 $0.50 $0.59 $0.44 $0.42 $0.52 18% Turkey $0.29 $0.36 $0.66 $0.26 $0.33 $0.46 29% Turkmenistan $0.59 $0.51 $0.94 $0.65 $0.55 $0.80 31% Tuvalu $0.79 $1.03 $1.05 $0.86 $0.91 $0.96 5% Ukraine $0.29 $0.40 $0.63 $0.20 $0.34 $0.42 17% United Kingdom $0.08 $0.15 $0.19 $0.05 $0.11 $0.12 5% Uruguay $0.61 $0.55 $0.94 $0.62 $0.58 $0.78 26% Uzbekistan $0.59 $0.54 $0.90 $0.35 $0.56 $0.63 10% Venezuela $0.22 $0.44 $0.50 $0.26 $0.33 $0.38 14% Vietnam $0.89 $1.24 $1.35 $0.97 $1.07 $1.16 8% Yemen $0.71 $0.96 $1.06 $0.86 $0.83 $0.96 13% Zaire $0.58 $0.75 $0.89 $0.66 $0.66 $0.78 15% Zambia $0.68 $0.57 $1.07 $0.69 $0.62 $0.88 29% Zimbabwe $0.29 $0.44 $0.74 $0.37 $0.36 $0.56 35%
Mean 25% Std Dev. 13% Min 1% Max 98%
61
Bibliography 1. Clark, D. David. “A Model for Cost Allocation and Pricing in the Internet”, Presented at MIT
Workshop on Internet Economics March 1995.
2. Mackie-Mason & Varian, H., 1995, "Pricing the Internet," Public Access to the Internet, ed. Kahin &
Keller, Englewood Cliffs, NJ: Prentice Hall.
3. Varian, Hal R. “Some FAQ’s about usage based pricing”, University of Michigan, September 1994.
4. Xie. J & Sirbu M., 1985, "Price Competition and Compatibility in the Presence of Positive Demand
Externalities," Management Science, June 1995.
5. Tracy, Lenore. “Internet Telephony: Who is Buying”, available at www.infoexpress.com.
6. Burkhard Stiller, George Fankhauser, Bernhard Plattner, Nathalie Weiler. “Charging and Accounting
for Integrated Internet Services- State of the Art, Problems, and Trends “, Computer Engineering and
Networks Laboratory TIK, Swiss Federal Institute of Technology, ETH Zürich, Gloriastrasse 35, CH-
8092 Zürich, Switzerland.
7. Francis Chow, Ingrid Ma, Ramon San Andres, Jun Shu and Kristin Wright, “Network Pricing, Costs
and Settlements”, available at www.berkeley.edu, May, 1997.
8. Anonymous, “Elemedia Pricing Models”, available at www.elemedia.com, 2000.
9. Mitrabarun Sarkar. “An Assessment of Pricing Mechanisms for the Internet--A RegulatoryImperative”,
Presented at MIT Workshop on Internet Economics March 1995.
10. Jeffrey K, MacKie-Mason, Liam Murphy and John Murphy. “The Role of Responsive Pricing in
Internet”, Presented at MIT Workshop on Internet Economics March 1995.
11. Emir A. Mohammed. “The Growth of Internet Telephony: Legal and Policy Issue”, available at
www.firstmonday.com, 1999.
12. Lamb, Chris. “Packet Telephony Primer”, 3Com Technical Papers, available at www.3com.com.
13. Anonymous, “ Packet Telephony: Long Distance Service for ISPs”, Business case available at
www.cisco.com.
14. Anonymous, “An Introduction to Cisco Open Packet Telephony for Service Providers”, White paper
available at www.cisco.com.
15. Anonymous, “Internet Call Waiting Solution Overview”, Overview available at www.cisco.com.
16. Hu, Howard. “IP Telephony: Paving the Way for Enhanced Services” ,Technical Paper available at
www.cisco.com.
17. Anonymous. “Sector Report: Internet Service Providers”, available at www.indiainfoline.com.
18. Moron, Rick. “Getting Ready for IP Telephony”, A supplement to business communication review,
Nortel Networks.
19. Mcknight, Lee and Leida, Brett. “Internet Telephony: Cost, Pricing and Policy”, Chapter submitted to
the volume of selected papers from the1997 Telecommunications Policy Research Conference.
62
20. Mcknight, Lee. “Internet Telephony and Open Communication Policy”, Internet Telephony
Consortium, MIT, 1997.
21. Clark, David D. “A Taxonomy of Internet Telephony Appluications”, Internet Telephony Consortium,
MIT, 1997.
22. Leida, Brett A. “A Cost Model Of Internet Service Providers: Implications For Internet Telephony And
Yield Management” MS Thesis, MIT, 1998.
23. Ladid, Latiff. “IPv6- The New Generation Internet”, Ericsson Review No. 1, 2000.
24. McGarty, Terrence P. “The Internet Protocol (IP) and Global Telecommunications Transformation”,
presented at the Seminar on “Creative Destruction: Telecomm in Transition”, Tufts University,
Medford, MA, March 12, 1999.
25. Andrew Sears, 1996. "The Effect of I-Phone on the Long Distance Voice Market," revised October 4,
1996.
26. Percy Alan, “Understanding latency in IP Telephony”, available at www.soundingboard.com 1999.
27. Mine, Hilary, “The Evolution of internet Telephony”, white paper available on www.att.com,
september 1998.
28. Anonymous, “AT&T’s IP Telephony Initiative”, white paper available on www.att.com, september
1998.
29. Gill, Navjit. “Hot Line on the Net”, Businessworld, 1 May, 2000.
30. Anonymous, “Quality of Service for IP Voice over DSL”, Product Brief available at Nortel Network.
31. Yeung, Angela, “Internet Telephony in Hong Kong”, available at
www.tradeport.org/ts/countries/hongkong/isa/isar0017.htm, March 1999.
32. Breiter, Maria. “Internet Telephony in Russia”, available at
http://www.comptek.ru/telephony/itsp_list1.html, November, 1998.
33. Longtao, Wu. “IP Phone over Wireless Data Access Network”, available at
www.telecomn.com\focus4_20002.htm.
34. Rong, Zhang. “Current Status and Future of IP Telephone”, available at
www.telecomn.com\focus2_20002.htm.
35. Sunderland, Kim. “The 1999 Regulatory Outlook for IP Telephony”, posted on
www.soundingboard.com.
36. Semaria, Chuck. “Multiprotocol Label Switching: Enhancing Routing in the New Public Network”,
white paper available at www.juniper.net, Juniper Networks, September 1999.
37. Anonymous, “IP Telephony basics”, available at www.dialogic.com
38. Anonymous, “ThruQoS for VoIP”, available at www.iphighway.com