Issue No. 03 - May/June (2002 vol. 6)
It has been three years since IEEE Internet Computing last dedicated an issue to Internet telephony, and it is interesting to see how much has changed since then — and somewhat depressing to see how little has changed, as well.
Defining the Technology
Three years ago the IETF published RFC 2543, which defined the session initiation protocol. SIP is the IETF's take on the end-to-end model of IP telephony. It was seen as a more Internet-centric alternative to the ITU-T's H.323. This was in the middle of the Internet boom, and there were great hopes for a quick explosion in the use of IP telephony.
A year and a half ago, the IETF — working with the ITU-T — published the megaco protocol (known as H.248 in the ITU-T). This architectural model is even less end-to-end than H.323. It is designed to support the creation of IP-based phone switches that could mimic traditional phone switches — just the sort of thing that large enterprises and phone companies tend to like for operational and business-model reasons. This standard closed a gap that many in the traditional phone world felt existed in the IP telephony world, and so again, many expected (or at least hoped for) a significant upsurge in the use of IP-based telephony.
Just a few weeks ago, the IETF approved publication of an update to RFC 2543. The update does not define a new protocol but rather, cleans up the details of the old specification, particularly in security. The quick explosion in IP telephony has not yet happened. Indeed, it is still rare to find someone or some organization using IP-based telephony in anything other than an experiment.
With dozens of implementations under its belt, SIP is becoming a mature technology and is being deployed by a growing number of organizations. To be a bit facetious, however, we could say that the closest estimate for the extent of operational deployment for all types of IP-based telephony is still zero percent because, at most, a few hundred thousand of the more than a billion phones in the world are IP-based.
THE BOOM TO COME?
By all projections, the number of IP-based phones will grow significantly over the next few years. Indeed, if the proponents of third-generation cellular phone technology are anywhere near accurate, the number of IP-based 3G mobile phones could soon number a few hundred million. But I'm not holding my breath. IP-based telephony needs to do more than just work as well as and mimic the public switched telephone network (PSTN).
As several enterprise-based trials have shown, IP telephony can work quite well. But it faces a hard challenge before it gains anything approaching ubiquitous deployment: The people who make telephony-related decisions in organizations tend to be mired in traditional phone-think and often see the future only as a small increment on the present. For example, they "know" what phones are and what services they offer and thus have a hard time imagining what could be done with the flexibility offered by bringing IP to phones. If IP telephony is seen only as a different way to provide the current concept of phone service, or perhaps only a subset of what can be provided today with standard equipment, then there is little incentive to deploy.
Traditional telephony services have been slowly developed over decades — their characteristics heavily constrained by a limited 12-button user interface. Imagining what phone service could be if the user interface were not so constrained has not been easy, but developments will come in time. Until then, only adventurous telco managers will be likely to do more than small experiments with IP telephony.
In This Issue
The articles that follow examine some of the other things that need more work before IP telephony becomes less experimental and more mainstream. In "Providing Emergency Services in Internet Telephony," Henning Schulzrinne (guest editor of IC's 1999 IP telephony issue) and Knarig Arabshian take a look at what functions IP telephony will have to support to provide telco-style emergency services over the Internet. They examine both everyday access to emergency services such as 911 and emergency alerts in the aftermath of network disruptions caused by natural disasters or man-made attacks.
One of the most common worries about IP telephony is the quality of the service offered. Most people's experience with IP telephony has shown a wide range of quality, from analog cell phone (via satellite link) on a bad sunspot day to better than any traditional phone. The next two articles explore aspects of the quality-of-service issue.
In "Assessing Voice Quality in Packet-Based Telephony," Jan Janssen et al. report on the impact of transmission parameters, and various types of network-induced errors, on the perceived quality of voice service. Starting with an introduction of the salient features of packet-based telephony, they then explore the impact of codec choice and network-error handling on the user experience.
The lack of reliable bounds on end-to-end delay can have a major impact on the quality of phone calls carried over the Internet. Hugh Melvin and Liam Murphy, in "Time Synchronization for VoIP Quality of Service," propose using synchronized time to improve service quality predictability in IP-based telephony. They show how time synchronization can help address this problem and discuss how they are testing the concept in a testbed network overlaid on the Internet.
Finally, in "Integrating Internet Telephony Services," Wenyu Jiang et al. describe the SIP-based Cinema system and the lessons learned when integrating IP telephony services with a PBX phone system at Columbia University. While the system has potential for use in many environments, it is also proof of the fact that only adventurous administrators are currently deploying IP telephony-based systems.
I do think that we have the basic technology well defined at this point, although we still need to work on some of the aspects pointed out by the articles in this issue. With the many trials now under way, we might find the need for additional work in a few other areas as well.
I think the main things standing between where we are now and meaningful deployment of IP-based telephony are no longer technical. There is a significant mind-set problem to be overcome with some traditional telco managers, and a set of viable business models for carrier-based IP-telephony have to be developed and proven.
If we can figure out ways to overcome these hurdles, we might find, when it comes time for the next special issue on IP telephony, that the topic is no longer all that special. Instead, it will be just part of the multifaceted service platform and innovation engine that we will still call the Internet.
Scott Bradner is senior technical consultant at the Harvard University Office of the Assistant Provost for Information Systems. He is a member of the Internet Engineering Steering Group, trustee and secretary of the board for the American Registry for Internet Numbers, vice president for standards for the Internet Society, and a member of the IEEE and ACM.