PDF of this tutorialHistorical Perspective
The packet voice revolution has slowly but steadily been gaining momentum over the last few years. The earliest deployments of packet voice were driven by enterprises using wide-area data networks to carry intraenterprise voice traffic as a cost-saving measure. The success of this approach was, however, short-lived because long-distance service providers brought in dramatically lower prices for virtual private voice networks, all but eliminating the cost advantages of packet voice on private data networks.
Internet telephony has been another key area of application for packet voice. The flat-rate access charges for Internet communications offer opportunities for major cost savings on long-distance and international calls. However, the relatively poor performance of the Internet in supporting real-time voice communications has limited its success as an alternative to the traditional public switched telephone network (PSTN)—apart from its potential to carry fax traffic, which is largely insensitive to transmission delays.
Outlook
Despite the patchy success of packet voice to date, market interest in this area remains very high. The primary reason for this interest is the dramatic growth of Internet protocol (IP) traffic relative to voice traffic in networks all over the world. The bandwidth consumed by voice traffic in the PSTN is growing at a slower rate than data bandwidth consumption growing in the Internet. If Internet traffic continues to grow at the same rapid pace, in a few year's time the total volume of voice traffic worldwide will be dwarfed by the volume of IP traffic, most of which is Internet-related.
Growth in the volume of IP traffic is driving demand for extremely rapid advances in both transmission and packet switching technologies. Dense wavelength division multiplexing (DWDM) techniques are squeezing multi-Gigabit capacities out of individual optical fibers, while core network routers are matching this growth in transmission capacity with multi-Gigabit routing capabilities.
As a result of these continuing advances, transmission and switching costs for IP traffic are falling fast. It is, therefore, reasonable to assume that voice transmission and switching costs will also fall quickly if voice were carried in packet form. Since there is no comparable innovation taking place in the circuit-switching world, one may also conclude that voice transmission and switching costs will not fall as quickly if voice stays on the traditional circuit-switched network.
This vision is driving the rapid developments that are now taking place in packet voice—one of which, of course, is the emergence of VoDSL solutions for packet voice access.
It is tempting to conclude that, in a world dominated by IP, voice will ride almost for free on this high-capacity IP infrastructure. But the requirements for voice transport are fundamentally different than those for data, and it is unclear at this point how a pure IP–based solution will be able to provide guaranteed delivery of voice packet streams with acceptable end-to-end transmission time on a global scale. This is a major concern of service providers because it is voice traffic that continues to dominate both their revenues and their margins. As a result, most of the IP traffic in these high-capacity packet networks is actually being carried over ATM because ATM offers common packet-based transport facilities for both voice and data while providing the quality of service (QoS) guarantees that are so important to voice.
The economic case for applying packet-switching techniques to voice is very clear. What is less clear right now is which packet switching technique will dominate the voice world in the long term. The proponents of IP–based voice may argue for the elimination of the ATM layer altogether; but until IP has demonstrated a clear ability to provide a scalable and economically attractive solution that protects the quality of voice transmissions, the safe choice for voice will continue to be ATM.
