Internet Telephony
How IP
Phones Home

by Isao Arima

According to market researcher IDC, at the end of 1996 there were a half-million active Internet telephony users worldwide. IDC predicts that this number will grow to 16 million by 2000, and 40 million by 2005, driven mainly by business users.

Internet telephony services can be classified into three types, as shown in the figure on page 25: voice communications between audio devices attached to computers, between a computer and a telephone, and between telephone terminals via the Internet.

The first type - voice communication between two PCs - can be accomplished by simply connecting two PCs running Internet telephony software via the Internet. This enables not only voice communication, but also such business functions as "shared whiteboard," document sharing, file transfer, and video conferencing.

Voice communications between a computer running Internet phone software and an ordinary telephone - the second type - is made possible by an Internet telephony gateway placed between a PSTN and the Internet. The caller first contacts an Internet telephony gateway, then follows the gateway's voice guidance message to input a pre-designated identification number. After authentication, the caller's connection is routed via the Internet to a telephony gateway near the receiver's location.

The third type of Internet telephony service - that between two telephones - can be implemented by connecting two local PSTNs via Internet telephony gateways. This application of the technology can offer the caller lower phone rates since it uses the Internet for the long-distance portion of the call.

By using a corporate intranet instead of the Internet, this type of service also can be used to support office-to-office voice communication. Connecting the offices' PBXs (private branch exchanges) to an intranet provides a cost-effective business application, since both the intranet data and PBX voice signals can be sent over a single leased line.

The issue of quality
The principle of voice transmission over the Internet is basically the same as that of conventional telephone services over a PSTN. Analog voice signals are encoded into digital data, and that data is then transmitted through nodes on the network and decoded back into analog signals at the receiving end. The big difference - and the reason for the disparity in voice quality - lies in the structure of the two networks.

A PSTN is a connection-type network: once the connection path of a call is established, communication quality is guaranteed. The Internet, on the other hand, is a connection-less (or "best effort") network: the voice data is stored in packets that are sent to the destination point via routers on the network. Not all packets will necessarily go via the same path, and there is no guarantee of the packet transmission reliability. Since some of the data packets may be lost or delayed during transmission, "clipping" of voice data can occur.

Several things can be done to improve effective voice transmission over the Internet. With a PSTN, once the connection is established, that bandwidth is available exclusively for that particular call. Internet bandwidth, however, is limited (and shared by many users), so reducing the data bit rate by compression is a useful improvement technique. And placing buffers on the network can help prevent the clipping of data. But while buffers can sort out data in the middle of transmission, they also serve to increase the transmission delay; the larger the buffers, the longer the delay. This degrades the quality and utility of voice signals. (Also, buffers are of no use for data larger than the buffer size, nor can they compensate for packets lost during transmission.)

Voice compression technologies
There are several types of voice coding/decoding (CODEC) technologies. With PSTNs, the most popular method is the "µ law" with a 64K-bps bit rate. Internet telephony, however, generally relies on some form of CELP (Code Excited Linear Prediction). CELP methods (also used for cellular phone communications) have a bit rate of 5K to 8K bps, although some Internet telephony systems use GSM 6.10, a European standard with a 13K-bps bit rate, or even RT24, with a 2.4K-bps bit rate.

In general, the slower the bit rate, the lower the voice communication quality - which is another reason that Internet telephony does not match PSTN service. However, since these methods do provide an audible level of voice transmission (on par with the voice quality often experienced in cellular phone conversations), many people are willing to accept the lower voice quality of Internet telephony as a tradeoff for the lower price.

Standardization and interoperability
For Internet telephony to become truly useful and widespread, the various systems in use must be able to connect with one another. Standardization activities for interoperability between Internet telephony systems were started in 1996 by the International Telecommunications Union (ITU), when it established a recommendation (ITU-T H.323) that addresses the methods of call control, multimedia management, and bandwidth management. This standard is applicable for multimedia communications using IP (Internet protocol) and supports all of the CODEC methods recognized by the ITU.

In addition to the ITU activity, a nonprofit industry consortium, the International Multimedia Teleconferencing Consortium (IMTC), has been involved in formulation of an industry standard for multimedia teleconferencing. Through the IMTC's efforts, a Voice over IP (VoIP) Forum was created. The VoIP Forum decided in March of this year to recommend the CCITT G.723.1 method as the industry-standard, baseline technology for Internet telephony.

In accordance with these standardization efforts, many vendors are now developing Internet telephony systems that use H.323 and G.723.1 CODEC. It seems certain that interoperability between competing Internet telephony systems will be advanced based on these. Regarding a directory service to provide an IP address search based on a person's name (i.e., to look up an individual's "phone number"), however, standardization activities are, so far, little advanced.

The lack of a "quality of service" guarantee for IP network-based voice communication remains a major stumbling block. To improve the situation, however, an industrial association, the Internet Engineering Task Force (IETF), has set up a protocol known as RSVP (Resource reSerVation Protocol). RSVP provides a receiver-initiated setup of resource reservations for multicast or unicast data flows in order to guarantee a certain bandwidth for end-to-end communications. While some providers have already started using RSVP for their commercial services, the protocol still has limitations. For example, RSVP has to be implemented in all of the routers of the communication path, and it cannot always secure the needed bandwidth (particularly when there are multiple requests for bandwidth guarantee at a time). It will be very difficult to provide a bandwidth-use guarantee for Internet voice communications unless there is a drastic refurbishment of the network.

CTI and Internet telephony
Computer Telephony Integration (CTI) provides intelligent control of phone calls by using computers. A simple CTI configuration consists of a PC, a modem, and a phone. [For more on this, see "CTI: Joining the Best of Two Technologies" in our June 1997 issue, page 17.-Ed.]

Also, as a corporate application, CTI provides unified messaging functions for comprehensive handling of e-mail, fax, and voice mail. CTI development is now proceeding for advanced connection of office phone systems with groupware, Internet, and mobile computing environments. Current CTI systems handle voice/data and control signals separately - phone lines for voice and data, and LAN (local area network) cables for signal control.

An Internet telephony system, on the other hand, can use an office LAN cable for both voice/data signals and control signals. There is a move toward convergence of technologies, with an Internet telephony system viewed as a new component of CTI. In the quest for unified messaging service applications, Internet and intranet resources could be shared, leading to improved business efficiency. You could access your office voice mail, for example, remotely via the Internet, while an urgent call coming to your office phone could be forwarded via the Internet to your (Internet telephone-equipped) laptop PC.

Toward the future
This article has reviewed some of the ways in which the technology to realize convergence between computers and telephony systems is being advanced for future growth of business opportunities. The regulatory issues impacting the deployment of this newly emerging technology into practical business applications, however, remain unclear. (For example, should the Internet telephony business be regulated as a telephony business or as a data communications business?) If Internet telephony technology is to be soon implemented into practical business applications, swift settlement of the remaining regulatory issues is essential.

After earning a master's degree in physical electronics at Tokyo Institute of Technology, Isao Arima joined NTT Data in 1990. From 1992 to 1993, he was dispatched to SRI International in the US as a visiting researcher for speech recognition and voice data processing technology. Currently, he is in charge of development of speech recognition technologies at NTT Data's Laboratory for Information Technology.

translated/adapted by Noriko Takezaki

Back to the table of contents