Reach Out and Net Someone

A few months ago Robert Howe, a professor of mechanical engineering at Harvard, tried to make a telephone call over the Internet. “It’s not something I’m likely to repeat in the near future,” he complains. “Ever experience the slight delay in an overseas call? Well, Internet telephony is 10 times worse.” The sound transmission was so choppy and unreliable, he says, that both parties became frustrated. “The delays threw off the cues you normally rely on in a conversation. It was more like typing than talking.”

Such problems are typical of attempts to transmit phone or videoconference communication over the Internet. Making the call is simple enough: it requires only some inexpensive software for both parties. But because digitized speech involves two to four times as many bytes per second as written text-and because voice communication is highly sensitive to even minor delays-telephony taxes the resources and current structure of the Internet.

In response to this problem, Lee McKnight, associate director of MIT’s Research Program on Communications Policy, in the Center for Technology, Policy, and Industrial Development, has organized a broad, international study group to reexamine Internet protocols-the rules and procedures that govern how data are transmitted. Members of the Internet Telephony Interoperability Consortium include Sprint, U.S. Robotics, Lucent Technologies, and a number of Japanese and European telecommunications companies.

Unlike conventional telephone service, phone communication over the Internet is free to anyone with access to the Net and the right software. Interest in Internet telephony is therefore high-but so far the actual number of users is low. Though Internet telephony is “not ready for prime time,” Knight says, his group is planning ahead to ensure that the Internet will be able to accommodate the demands imposed as the use of telephony increases.

The poor quality of Internet telephony is a byproduct of the way the Net transmits information. Internet protocols break messages down into manageable “packets” of information, which move from computer to computer along the Net and reassemble at the point of delivery.

According to Andrew Sears, an MIT graduate student in electrical engineering, computer science, and technology policy who is working with McKnight’s group, three potential bottlenecks can choke this process when telephony is involved. First, your own computer must digitize voice transmissions by employing data-compression software, causing delays of a few thousandths of a second. (Operating in reverse, this process is used to reassemble the message at the receiving end.) Second, if the volume of data to be transmitted overwhelms the available capacity, or bandwidth, long lines can form at each juncture as packets wait to be routed from one computer to the next. Thus hundreds of opportunities for delay can occur in a single transmission. Third, because the packets that compose a single message travel different routes to their final destination, chunks of information may become lost or delayed-taking the long way around, so to speak.

Such delays cause no more than minor inconvenience in transmitting written text, but delays in spoken communication make conversation frustrating and even unintelligible. “Right now,” McKnight says, the Internet offers “no quality guarantees. That’s one reason why it’s so cheap to use. If it gets congested, it discards or delays information regardless of content or user preference.”

One solution his group is studying is reserving bandwidth for telephone or videoconference calls-the equivalent of setting aside a high-speed lane on the information highway. The consortium is investigating a variety of pricing mechanisms through which consumers could purchase access to this reserved bandwidth. RSVP, one of numerous protocols proposed by McKnight’s group, would enable e-mail aficionados to maintain their inexpensive service while other users pay extra to subscribe to voice-quality or video-quality Internet service. Other options the consortium is investigating include allowing consumers to pay a priority rate for each phone or video message or to reserve access for high-priority messages at particular times or for calls to particular phone numbers. Even with these higher rates, Internet telephony is likely to remain cheaper than the conventional alternatives.

Students working with the consortium are developing business models to help Internet access providers identify potential markets and redefine their services accordingly. According to Brett Leida, an MIT graduate student in electrical engineering and technology policy, if telephony accounted for one-third of the time users spend online, the costs of providing Internet access would nearly double, since companies that provide the network of fibers along which information is transmitted would have to build or lease more capacity and smaller companies that provide consumers access to this backbone would have to invest in capital equipment, particularly modems. (Modems account for about 70 percent of small providers’ capital investment. Because telephony users are likely to dial in more frequently or stay online for longer periods than other users, explains Leida, companies that provide Internet access to dial-in users will need to invest in additional modems to handle the incoming data stream.)

Many smaller providers, Leida says, are concerned about the potential cost burden because it is difficult to raise rates in this competitive market. One potential strategy is to shift their customer base from dial-in users (households and small offices) to larger businesses with whom they can maintain direct network ties, precluding the need to invest in modems.

Apprehension over these growing costs has also fueled the interest of providers in user-differentiated pricing, Leida notes. Offering a less expensive rate to users who make Internet phone calls only during certain hours could smooth out peaks in demand, offsetting the pressure to invest in additional modems.

McKnight’s group also plans to address policy questions such as the role of the Federal Communications Commission in regulating Internet telephony, and the potential for offering universal access to Internet telephony as an alternative to conventional telephone service. “Telephony places many new demands on the Internet,” he notes. “We want to be ready with suggestions and models to prepare the way.”