A Revolution in Voice Networks – VoIP

By Jon Kenton & Philippe Chevallier – Motorola Computer Group

Introduction

For many years the networks that have been in use to transmit our telephone conversations have been
based on the same basic technology, switched circuits. These networks have evolved many times
including one major evolution from analogue to digital circuits. We are currently however at the onset of a
major Revolution that is sweeping across the telecom industry and bringing in many new players who
hope to take advantage of a movement to transmit voice calls over the internet. We are all now involved
one way or the other in the “packet revolution”. It is called many things, Voice over Packet, Voice on the
Net, Internet Telephony or Voice over IP (Internet Protocol) or VoIP.

This revolution, brought about by the explosive growth of the Internet, is driving a major shift in the
spending of the infrastructure dollar towards high-speed packet technologies. It has also made it apparent
that the days of shoehorning data onto the voice network are ending and that it is time to start adapting
voice traffic to the Internet.

History

Since its inception the telecommunications network has been based upon circuit switching and for good
reason. Until the 1970s the only traffic on the network was voice telephony and the transmission of voice
telephony has essentially been based on creating a physical path (or circuit) between two handsets by
closing a series of relays or switches. From the days when an operator manually patched you through
their mechanical switchboard, to the present when your phone call is automatically switched through
enormously expensive high-speed digital switches, the phone system infrastructure was designed for a
single function, circuit switched voice. In the 1960s however, with the advent of commercial computers,
these same phone switches began to be used for data communications albeit awkwardly. The only
solution was the analog modem. Over the years the analog modem overlaid data onto the phone network
at ever-increasing speeds but has always been limited by the basic issue of bandwidth. This is primarily
because the phone network has been limited to audio frequencies from 300Hz to 3400Hz and is also very
“noisy”. Although people can make out a voice conversation through certain levels of static a modem has
a more difficult job.

In the early 1980s the telephone networks evolved into a purely digital system. From Central Office (CO)
to CO the analog voice signal was encoded as a digital stream and then switched digitally until the stream
reached its destination CO switch where it was converted back to analog. The connection from the CO to
the home (the local loop) was still basically an analogue two-wire connection. This worked reasonably
well until increasing loads were put upon the switching infrastructure by dialup data from a new and
rapidly growing phenomenon, the Internet.

The Internet

The Internets growth and popularity surprised everybody especially the capacity planners for the regional
network operators. Nobody truly anticipated the explosive growth in bandwidth demand and the overall
effect on the circuit switched infrastructure. While the Internet itself is based on packet switched
architecture, private individuals and businesses would access it almost exclusively via modems and the
established circuit switched local loop. The usage model developed by the telco capacity planners was
based on the typical length of a voice call, which was in the order of minutes. With US local calls being
free, and ISPs offering inexpensive “all you can eat” Internet service, these dialup connections were often
“offhook” for hours. Some people even left them on 24 hours a day. This severely impacted the capacity
for regular voice calls tying up expensive ports at the switching offices.
 1800
1600
1400
1200
1000 Voice
800 Data

600
400
200
0
1998 1999 2000 2001 2002 2003

Fig 1: Voice growth vs Data growth

The new network solution

As more and more people hooked up to “the net” the issue was exacerbated and it continues to escalate.
In the U.S it is estimated that total data bandwidth will have exceed total voice bandwidth by the year
2002 (figure 1.). The solution to these problems has been the evolution of a new digital packet switched
network that can route the data traffic separately and keep it away from the circuit switched voice
network.

The local access problems now have many solutions also, with the advent of Cable modems, satellite
connections and DSL (digital subscriber line) services. These technologies all have the benefits of higher
speeds often in the 1-5Mbps range as well as direct access straight into the Internet without ever having
to touch the conventional telephone network.

Voice on the Internet

So now we have two distinct networks one for voice and one for data. As well as the fundamental
circuit/packet differences the billing structure is very different indeed. We are used to paying more the
further we want to call especially for international connections. On the Internet there is no distance
component, once we pay for our connection into the net it costs the same to access a site in the next
building as it does to get to one clear across the globe.

This caused many to think that with data having been carried on the voice networks for so many years,
why not put the voice on the data network. This has lead to many opportunities where new companies
have setup operations as voice providers when they own only a packet switched data infrastructure and
build interconnections into the conventional phone networks. Service revenues from VoIP have been
small by comparison to existing services but are forecast to exceed the $5billion mark by 2003.

A whole new architecture has been built and continues to evolve to support the Internet telephony
revolution. As you would expect there are many hardware and software components, the next section
provides a breakdown of the various architectural components and how they all fit together. The growth in
sales of these key components is another indication of how rapidly this new infrastructure is taking hold
and will begin to start taking service revenues from the existing networks. Figure 2 shows the growth of
VoIP gateways which are responsible for the interconnect of the old and new networks and according to
the MMTA have a CAGR approaching 40%.
 1600
1400

Fig 2: 1200
VoIP Gateways 1000
$M U.S. 800
Growth 1996 – 2003 600
400
200
0
Source MMTA 1996 1997 1998 1999 2000 2001 2002 2003

The fundamental building block components for VoIP

The term Voice over IP refers to a delivery mechanism that allows any IP based network to carry voice
with higher or lower quality depending upon how the network is being engineered. On a dedicated Local
Area Network (LAN) for instance, the voice quality can be higher than on the regular Public Switched
Telephone Network (PSTN). However, if a network should become saturated, compression mechanisms
become very important to maintain QoS (Quality of Service) levels. To communicate over an IP network,
signaling protocols similar to the ones used by the PSTN such as SS7 or Channel 7 have been created.
To setup, tear down and communicate over an IP network, VoIP uses four different types of elements:

1. Terminals
2. Gateways
3. Gatekeepers
4. Multipoint Control Units.

All these components have a different role within the network, although some of them are optional such
as the gatekeeper, and all can be either built into a single system or spread over multiple systems at
different physical and geographic locations.
 Terminal Gatekeeper Front end Front end Gatekeeper
Terminal

IP Network
Video - H.320 Video

PSTN
Media Media
Data - T.120 Gateway Gateway Data

IP Phone - H.324 IP Phone

Telephone ISDN
Multipoint Control Unit
Video - H.320

H.323 Network model

Terminals
A Voice over IP Terminal or client is the communications element that helps connect to real live calls.
There are a number of client types found today and they must all be capable of supporting voice
communications at a minimum but can optionally provide support for video or even data communications.
The primary terminal type in use today is a software package such as Microsoft's NetMeeting, running on
a PC. This presents the user with an interface through which they can make calls across the Internet. In
the background it is responsible for the setup and teardown of the calls as well as the encoding and
decoding of the transmitted/received voice such that the PCs microphone and speakers can be used
instead of a regular handset. If you wish to use the PSTN and a regular telephone to make use of the
Internet for voice calls a “virtual terminal” is used. This is a software module that would be part of a VoIP
gateway deployed by the service provider and this provides the interfaces & protocol endpoint for the
communication. This would then be converted by further functions inside the gateway into a regular
phone connection.

So the terminal is the end user service that provi des real time, two-way voice, video or data
communications with other VoIP terminals. It communicates with the VoIP gateways using H.245 for call
control, Q.931 for call setup and RAS for registration and administration with its local Gatekeeper. It
interfaces with various elements such as a regular Plain Old Telephone Set (POTS) or
microphone/speaker for the audio side, or to a camera/monitor for video transmission.
 Screen

Video Telephone
Television

Audio Codec
Video Codec
G.711 G.723
Data H.261/H.263
H.245

Q.931

H225
G.729

RAS
Interface
T-120
RTP/RTCP

TCP UDP

IP

H.323 Terminal Software Model

Gateways
The whole concept of VoIP could not be totally viable if Internet telephony users could not talk to regular
telephones. VoIP Gateways, provide the interconnection between the old school of traditional telephony
and the digital world of Internet telephony. This way users of both technologies are able to communicate
with one another. The primary function of the Gateway device is to provide translation services for the
previously mentioned ‘virtual’ terminal as well as different transmission formats, communications
procedures, and audio codecs.

The Gateway is a two-way interface between the telephone network and the IP-based network, therefore
optional when there is no need to interconnect with a regular public switched telephone network, such as
in an enterprise LAN only architecture. It assumes responsibility for setting up and taking down voice
channels between H.323 and the PSTN network such as T1, B-ISDN, SS7, etc.

The Gateway model can be a single box that provides this interface but the Gateway model can be
decomposed into three separate components and running on three different platforms:

Gatekeeper
RAS

SIGTRAN
SS7
H.245 SS7 - Link
Media Gateway Signaling
Controller Gateway
H.323 Terminal MGCP / Megaco
/H.248
PSTN
RTP Voice/ TDM - Voice/
Video traffic Media Video Traffic
Gateway Media Gateway

H.323 Gateways
Network model
1. Media Gateway: provide the voice traffic translation between an IP based G.723.1 at 6.3kbps to
G.711at 64kbps. From one side it is connected to a Local Area Network such as Ethernet 10/100BT,
 and on the other side it assumes the connection to the telephone network as a T1 trunk or ISDN line
for video communication with H.320 compliant video equipment. This platform is required to remain
active all the time to prevent any discontinuity of service between two end points. A High availability
platform is required with a minimum downtime and allows maintenance operation while the system is
operational. This node controls jitter, delay, echo cancellation, or any other component that constitute
quality of service (QoS).
2. Media Gateway Controller: provide the overall control of the gateway. It communicates with the
Gatekeeper for database information regarding mapping between IP address and phone network.
3. Signaling gateway: responsible for the interface between SS7 signaling network and the VoIP
signaling such as H.323.

Gatekeepers

Call control is key to any network managing voice communications; Gatekeepers provide these functions
within the IP network. Many of these functions are provided by complex database management systems
and include billing, address translation, routing and bandwidth management.
Database software constitutes the primary element of this platform. However, due to the kind of
information this platform is required to provide, a fault tolerant or high availability platform is required with
a reliable database management environment. Gatekeepers are linked together using a “border-element”
or Super-Gatekeeper using the H.225 Annex G protocol definition.

Address Admission Bandwidth Zone

translation Control Control management

H.225

TCP/IP

Gatekeeper S/W modules

Gatekeeper mandatory functions

1. Address translation
This function allows any endpoint to retrieve a transport address from an alias address or vice versa.
On an IP network, this feature prevents H.323 endpoints (terminal, gateway, etc..) from keeping the
alias to IP address translation locally, therefore preventing connection errors or unknown IP address.
2. Admission Control
The Gatekeeper provides connection admission control based upon network bandwidth availability,
authorization, or any other criteria that is relevant for a specific implementation.
3. Bandwidth Control
The gatekeeper manages the bandwidth over the network and allows the best communication quality
among endpoints.
4. Zone Management
Defines which H.323 endpoints are currently managed by this Gatekeeper. Each endpoint is
responsible for requesting a registration to this module, therefore benefiting from all the features
provided by the Gatekeeper.
Overall, the Gatekeeper is a simple heavy -duty computer with high availability characteristics.

Multipoint Control Unit

One of the Internet’s inherent advantages is the ability to build cooperative and collaborative
environments. This ability is used heavily in the enterprise. The MCU is the component of the VoIP
architecture that allows users to further collaborate by taking part in either telephone or videoconferences.

The MCU acts as an endpoint in the network that provides the capability for three or more H.323
terminals to participate in a conference. It consists of two parts: a Multipoint Controller and a Multipoint
Processor (optional). The gatekeeper can explicitly invoke the MCU when more two or more endpoints
are participating to the same conference call.

Multipoint
processor
H.245

RTP/RTCP

TCP UDP

IP

MCU

The MP provides the mixing and switching of all the audio/video and data among all the H.323 endpoints
under the H.245 control. It has the same audio/video Vocoder functions as included into all other H.323
terminals or gateways, therefore allowing all automatic call attendant functions. For instance, it is possible
to send a text to speech message to all attendants, or provide a DTMF function for specific features that
the MCU would provide. Since the MCU has no geographic location requirement and since it uses the
same element as other pieces of the H.323 network, it can be located on a local Gateway or Gatekeeper.

Standards and Protocols

There are two active groups working on the standards for VoIP. They are the ITU-T (International
Telecommunication Union) Study Group 16, and the IETF (Internet Engineering Task Force). The IETF
has a number of groups working on different aspects of IP telephony, including the MMUSIC (Multiparty
Multimedia Session Control) working group. Two major standards are being recommended and deployed
to satisfy the same needs H.323 (ITU) & SIP (IETF). SIP encoding is based on textual formats such as
HTML or SMTP, etc and while simpler is not currently as widely employed as the H.323 standard. H.323
is based more on a traditional protocol stack with complex encoding and layering e.g. ASN1 within the
traditional 7 layer OSI stack. The OSI approach had been limited by its cumbersome size and the
compute power needed for such complex processing. With today’s technology this is no longer a
limitation that could hamper success of H.323 based solutions.

??ITU-T - H.323
The ITU recommendation was ratified in its first version in 1996 and approved in its version 2 in
February 1998. This recommendation describes the multimedia communication mechanism over a
 network with no quality of servi ce. Due to its latency and, non-deterministic aspect, an IP or IPX
network over Ethernet, Fast Ethernet or token ring is considered to constitute such a network. H.323
recommendation uses the following protocols and recommendations:

?? RTP: (Real-time Transport Protocol) provides end-to-end delivery services of real-time audio and
video. It provides payload-type identification, sequence numbering, times-tamping, and delivery
monitoring.

?? RTCP: (Real-time Transport Control Protocol) is the counterpart of RTP that provides control
services. It provides feedback on the quality of payload. Other functions include carrying transport
level identifier for synchronization of audio and video data.

?? H.225.0 RAS: Registration, Administration and Status. This layer is used between the IP
Gateway and the IP Gatekeeper to exchange call setup and call tear down information, IP
address mapping, etc…

?? H.225.0 Call Signaling: used to establish a voice channel between two H.323 endpoints. The
call-signaling channel is opened between two H.323 endpoints or between and endpoint and the
gatekeeper.

?? H.245.0 Control Signaling: used to exchange end-to-end control messages governing the
operation of the H.323 endpoint. These control messages carry information related to the
following:

?? Capabilities exchange,

?? Opening and closing of logical channels used to carry media streams,

?? Flow control messages,

?? General commands and indication

?? H.323 Annex D: Real time fax over H.323 systems.

This recommendation allows the transmission of a fax over a packet based communication
system supporting H.323 standard. It describes the procedure for opening of a channel and the
sending of T.38 (fax format) packets as would normally be done between two physical or logical
fax endpoints.

?? H.450.x: Supplementary services for H.323. These recommendations specify signaling

procedures for optional features that H.323 endpoints (terminal, media gateway) may support.
These services are fully deployed and used by today’s telephone infrastructure.

??H.450.2: Call transfer

??H.450.3: Call diversion

??H.450.4: Call hold

??H.450.5: Call park and call pick up

??H.450.6: Call waiting

??H.450.7: Message waiting indication

??IETF and SIP

 The IETF working group MMUSIC has developed several standards within Internet conferencing and
telephony. The ITU would seem to have had a head start but IETF has been working on developing
standards specifically with telephony in mind from the outset. The ITU approach considered the end
to end protocols first and then looked to add the telephony specifics. The primary ratified and
currently most utilized IETF standard is known as, SIP (Session Initialization Protocol) and is
published under RFC 2543.

Last year, the IETF spun off SIP development into an independent working group to concentrate
specifically on the work started within MMUSIC. The SIP group still maintains a close relationship with
MMUSIC along with other related working groups such as Iptel (IP Telephony), Pint (PSTN Internet
Interworking) and DCS (Distributed Call Signaling).

SIP takes a very different approach to VoIP than the H.323 standard. It is more akin to a series of
recommendations than hard protocol boundaries. Even though its approach is simpler, in contrast to
the ITU’s H.323, additional services such as call transfer, call waiting, etc. are already included within
the initial recommendation.

DCS
Distributed Call
Signaling Group

Call Processing
IPTEL Language SDP (MIME) MMUSIC
IP Telephony Multiparty
Multimedia Session
Control

SIP
Working Group

Pint
PSTN Internet
Interworking

IETF Working Group relationship

When studying the differences between these two protocol sets lots of pros and cons can be identified
and only the future will tell which one will become the ultimate basis for the world’s VoIP environments.
The SIP versus H.323 issue looks not dissimilar to the old X.400 versus SMTP battle, and currently real
deployments seem to be in favor of H.323.

Other protocols are being developed by the IETF to communicate between the various pieces of the
Media Gateway.

??SIGTRAN
Part of the IETF, the SIGTRAN working group’s primary charter is to address the transport of packet
based PSTN signaling over IP Network. Within the VoIP architecture, it addresses more specifically
the communication protocol between the Media gateway and the Signaling gateway.

??MeGaCo & MGCP Media Gateway Controller Protocol

 Since the MG (Media Gateway) and the MGC (Media Gateway Controller) can be two distinct network
elements, these Protocols are used to setup and tear down voice channel based on the signaling
protocol (SS7, H.245, Q.931). However, this time the IETF and ITU-T (Study Group 16,) have agreed
to publish a combined effort between the two working groups. This will allow an unified media
gateway control protocol that is currently to be specified from the ground up. It will integrate all
services needed to support VoIP media gateways for audio, video and data communications as well
as packet networks such as ATM. Most of the large vendors such as Cisco, Lucent, Nortel, etc. have
already committed to support such a protocol. This recommendation will be published by the ITU
under the H.248 standard and is currently known as H.GCP.

Quality of Service

Among the most important issues that the IP telephony world has needed to resolve is that of Quality of
Service (QoS). We have come to rely on the existing Public Switched Telephone Network and expect
similar standards of service from any new networks. Due to the variable availability of bandwidth as well
as the networks possible loss of data, VoIP has to address issues such as audio quality, signaling and
call reliability in order to match the PSTN. Following is a description of the different elements involved in
this challenge.

?? Voice Compression: To minimize the amount of data transmitted over the packet network, the
voice payload is compressed on its way into the network and decompressed when it reaches its
destination. To perform this operation, special DSP (Digital Signal Processing) chipsets such as
those developed by Motorola and Texas Instruments, are being used. This voice compression
decompression functionality is also known as a “vocoder”. The issue related to the use of DSPs is
due to the trade off between the compression required to limit the amount of data to be
transmitted versus the audio quality. A high compression can result in a high degradation of the
voice quality. Also, during the compression/decompression operation, an additional delay other
than that caused by the network, is generated. To compare Vocoders, each algorithm is
associated with its MOS (Mean Opinion Score). MOS is a number from 0 to 5. The higher the
MOS is, the better the voice quality. For instance G.711 compression takes about 0.75ms and
has a MOS of 4.4. A G.723.1 has a MOS of 3.5 and a delay of 30ms, however G.711 needs
64kbps of bandwidth as opposed to 5.3kbps with G.723.1.

The following table captures three kinds of compression algorithms currently available on IP
telephony networks.

Compression Algorithm Data Rate MOS Score Delay (ms)

G.711 64kbps 4.4 0.75
G.726 32kbps 4.2 1
G.723 5.3kbps 3.5 30

?? Latency Delay: During a conversation between two phones or two endpoints, each physical
element constitutes a delay to the propagation of the payload. Such elements being the Gateway,
network, switches, DSPs, etc… This delay needs to be minimized so as not to become
significantly annoying and obviously noticeable to the user.
?? Jitter: Due to the IP based network, packets are not transmitted in a deterministic manner. If
packets are not being delivered at regular time intervals, a conversation could be bursty and not
very pleasant. To minimize this deficiency, a buffer is inserted at the receiving end that ensures
delivery of the payload at a constant data rate. This is called the Jitter. Jitter requires fine-tuning
and can be difficult to administer for each type of network and Jitter buffers only add to the
latency issues.
 ?? Echo cancellation: VoIP terminals can be subject to echoes caused by sound loops created
between a speaker and microphones located on that same unit. It is necessary to cancel this
echo first to minimize the amount of overhead this would generate to the network as well as the
discomfort to the conversation itself. Echo-canceller algorithms such as G.168 are inserted within
the IP network to eliminate such “noise”.
?? Throughput: Each voice channel requires a minimum network bandwidth. If the network does not
provide such bandwidth, the call can be terminated.
?? Packet Loss: During network congestion, some packets could be discarded this can cause
significant voice degradation and eventually loss of service.
?? Signaling quality: To establish a voice communication, it is important to provide fast and reliable
call setup and call tear down of a voice channel so a caller does not experience undue delays in
making calls.
?? Availability: A voice network is expected to be operational whenever anybody needs to use it.
We now take this as read when using the PSTN. VoIP networks need to meet this same
expectation.

To help quality of service within a network, working groups such as Diffserv (IETF) are developing
mechanism to improve the delays a network can generate. Methodologies are implemented to help
routers to speed up their decision algorithm, and minimize the overall time delay of such network. The
IETF has recommended a new protocol called MPLS (Multi-Protocol Label Switching) to help in this area.

What the future may hold

The IP telephony environment is required to provide the same type of services currently being offered by
any Public Switched Telephone Network. Such services include audio/video conferencing, 911, toll free
(800, 888), directory services, etc. Such services are as of today limited by the current VoIP
implementation. New standards and recommendation are being developed as listed bellow:

??IETF – SIP
The SIP working group is currently working on new capabilities based on the call control specification,
multiparty services and MIB support, etc…

??ITU - H.323
The ITU is also working on specifying new services to meet the current IP telephony trend as follows:
?? H.225.0 Annex G: Inter Domain Communication. Defines the protocol used to interconnect two
“Super-Gatekeepers” and share the telephony information between two IP networks.

?? H.323 Annex C: H.323 on ATM

?? H.323 Annex D1: Generic Functional Protocol

?? H.323 Annex D2: Call Transfer Supplementary service

?? H.323 Annex D3: Call Diversion Supplementary Service

?? H.323 Annex E “Multiplexed Call Signaling Transport” defines multiplexed transport layer. First
goal being UDP based H.323 with optional reliability methods such as

?? H.323 Annex F “Simple Endpoint Type”

?? H.341 “Multimedia MIB”

Regarding audio Vocoders, new DSP algorithms are being introduced to allow high quality audio
traffic. These type of algorithms are meant to be used for radio broadcast, teleconferencing, or any
other type of audio transfer that requires high quality audio on a low bandwidth network. One of them
 is known as G.722.1 and has been developed by PictureTel. It consists of a wideband audio coding at
50 to 7000Hz audio bandwidth as oppose to a traditional narrowband 300-3400Hz, and can be
transferred over the IP network 16, 24 or 32 kbps.

Hardware reliability and availability

Linked to QoS of the software and protocol environment is also the establishment of a reliable and highly
available hardware platform. In the world of conventional telephony we have come to expect “infinite
dialtone” where the phone always works no matter what. Voice over IP must be able to meet the same
standards of availability, however many of these new technologies are being built upon open standard
systems. The advantage is time to market, as the long development times of the older proprietary
telephony switch are untenable in this new envi ronment. As such many developers are looking to
technologies that can provide them the stable platform required on which to base their “carrier grade”
VoIP solutions. CompactPCI is proving itself as the open standard technology of choice in this area.

The technological benefits of CompactPCI are now well recognized. Growing out of the mainstream PCI
community the foundation is a solid one, based on tried and tested chipsets. The physical limitations of
PCI slot cards made them impossible to safely be included within designs requiring high levels of
availability and reliability. Coupling the PCI electrical specification with the IEEE 1101 mechanical
standards gives CompactPCI the robustness required for carrier grade infrastructure deployment. One of
the ultimate “killer benefits” provided by CompactPCI is its intrinsic ability to provide the hardware
fundamentals required for high availability architectures. This is due in no small part to the Hot Swap
specification which defines 3 models, Basic, Full and High Availability. The high availability model, while
inheriting the attributes of the others (physical hotswap and software control) adds the ability for the
application to control the system at a slot-by-slot level. This creates an environment where the software
can have ultimate control of the hardware, literally “turning off” a slot that may have a failing component.
This type of isolation is vital for the fault management functions required of a high availability solution
aiming for the 5NINES goal, which translates to only 5 minutes of planned, or unplanned downtime each
year.

At this time, Motorola Computer Group’s CPX8000 range of CompactPCI platforms is possibly the only
CompactPCI solution proving capable of delivering these levels of availability. The CPX8000 family of
CompactPCI® systems, when combined with appropriate software, is designed for critical telecom
infrastructure applications that must meet 5NINES (99.999 percent) availability. CPX8000 systems have
built-in redundancy for active system components—including system-slot central processing unit (CPU)
boards—enabling active modules to be exchanged for repair or upgrade while the system continues to
operate. Designed as a carrier-grade platform for operation in network equipment building standards
(NEBS) and European Telecommunications Standards Institute (ETSI) environments, the CPX8000
family is particularly well suited for switching applications and deployment within unattended sites.

Conclusion

The market for VoIP solutions continues to increase building upon the huge growth and expansion of the
Internet overall. This is underlined by manufacturers both established and startup who are rapidly
releasing products to meet the needs of the growing community of Internet voice service providers. The
standards will undoubtedly continue to evolve, adding features to match the existing infrastructure as well
as new ones that capitalize on the packet data foundation of the new networks. As dependence on this
new environment increases so will the demand for higher quality of service and highly available platforms.
While there have been many attempts in the past to introduce new revolutionary technologies many have
not be able to make the grade, VoIP has already established a significant foothold. The VoIP revolution
continues unabated, Voice on the Net is here to stay – watch this space.