This article was published in CAUSE/EFFECT journal, Volume 22 Number 2 1999. Copyright EDUCAUSE. See http://www.educause.edu/copyright for additional copyright information.
The Case for Plain Old Internet Telephony
by Jim WilliamsVoice over IP (VoIP) isn�t the same as Internet telephony. Most of the discussions among those of us associated with institutions that might have a PBX or a requirement to supply voice service to our users focus on VoIP. We tend to think of VoIP as a natural evolution, upgrade, or migration of existing voice systems. We tend to dismiss Internet telephony as something capable of replacing our current global voice architecture. Most of us who have tried it find it lacking. And it is easy to identify the deficiencies in both technical and human terms. But if the Internet is anything, it is about rapid change--more like revolution than evolution. The Internet is more about changing types and patterns of communications than evolving extant forms. Few would have predicted the substantial changes under way because of things like e-mail and Internet commerce. Internet telephony as we know it today may have appeal only to small niche markets, but the Internet may change all that. Consider the following.
Economics
Until very recently, most of the Internet, and most other data communications services for that matter, were built upon a network architecture designed and optimized for voice. Traditional voice systems assume "dumb" devices on the end with the "intelligence" for the service provided by the network--the PBX and the local and long distance exchange carriers. An intelligent network costs more than a dumb one.
Those orange tubes that continue to be planted along America�s right-of-ways do or will contain bundles of fiber. Less apparent perhaps are the latest generation of wave division multiplexers and big routers that transform light into something more useful. Even with today�s technology, each fiber is capable of carrying the equivalent of more than 10,000 simultaneous voice connections. While the technology required to make these fibers useful is expensive, it is much less expensive per unit of bandwidth than that available just a few years ago. The capacity of these networks has the potential to dramatically transform the economic architecture of telephony and other forms of communication. The bandwidth required to serve a voice call is insignificant compared to that required for other communications services that might enter your home or campus. It won�t be free, but it will likely become a decreasing fraction of your total communications costs.
Those who have invested in this new fiber and new technology still enjoy a pricing umbrella based on traditional technologies. Carriers all over the world continue to use DS0 (the lowest data service level), or some measure based on a circuit capable of carrying a single toll-quality voice call, as the fundamental unit for pricing circuits. Voice revenues still far exceed data although on many circuits the traffic is destined for a computer rather than a telephone. From a carrier perspective, that is a fortunate tradition. But ponder what might change as the cost to deliver bits continues to decrease by orders of magnitude.
Technology
The speed of light through a vacuum is constant. We know how to slow it down by passing it through other media, but we don�t know how to make it go any faster. All other issues, including the transmission time for electrons across a semiconductor interface, are open for improvement. The objection to today�s models for Internet telephony is largely because of delay or inconsistencies caused by the network or the devices on the ends. One of the large contributors to delay is the processing speed on the end device (the PC). Even though operating systems tend to increase in bulk, most of us would agree that time required to process voice decreases in successive generations of processors whether through brute force or discrete digital signal processing. The general purpose computer on your desktop is very capable of doing the analog to digital conversions required for Internet telephony. It just needs to do it faster to bring delay down to an acceptable level. Expect a considerable diminution in the nearly half-second of delay currently attributed to the PC.
The "best effort" Internet remains a problem. Response times on the commodity Internet seem to be improving. Transcontinental transit times of less than one-tenth of a second are not uncommon across the Internet. Services like cable modems and DSL are bringing such services to homes. On the horizon are quality of service (QoS) features like DiffServ. And the next generation of Internet protocols (IPv6) will contain a number of QoS features that will improve real-time applications like Internet telephony. Although not yet available on the commodity Internet, they are available on advanced networks like the vBNS and Abilene. Can Microsoft Windows be far behind? All of that, of course, is subject to change. In the meantime, the current Internet seems to be staying ahead of the curve and I suspect it will continue to improve.
Quality
Plain old telephone services (POTS) still have one basic function: enter some code (numbers) and exchange intelligible sounds with a person otherwise out of hearing distance. Sure, we add other features, but that is the primary objective--to be able to complete those exchanges when we want or need them. It works. The reliability of the telephone service in the United States may be beyond that achievable in at least the near future by the Internet, but replacing the existing telephone system with Internet telephony is probably not the right target. Price alone will not cause waves of consumers to shift to Internet telephony given the affordability of the current system. But that is not the Internet way. The Internet didn�t just replace phone calls to stock brokers, the ordering of books, and lookups in the yellow pages. It provided a new kind of service that wasn�t previously available. Here is one example of how that might apply to telephony.
That telephone on your desk, connected to whatever network, has a hard time pushing more than 3000 Hz of audio back and forth. That�s what we accept as a toll-quality, intelligible signal, but it is hardly CD quality. Compared to today�s Internet capacities, that circuit-switched voice call consumes relatively trivial amounts of bandwidth--especially if some of that carriage is over IP-enabled (VoIP) networks. But increasing that fidelity is such a difficult process that it is done only in special cases. Could Internet telephony do it better? You bet. If we all had just a little bit more reliable bandwidth, and maybe a little more processor speed, and someone clever to hack together a product, we�d have hi-fi, home-to-home, two-way communication. Perhaps stereo. Or surround sound. Maybe...
But then we probably wouldn�t call it Internet telephony.
Never mind.
Jim Williams ([email protected]) is director of policy analysis and federal relations at EDUCAUSE.