Today, the world has two separate and distinct public networks. The public switched telephone network (PSTN) has been in operation for well over a century. The technological advances have been impressive, but the principle is little changed from the days of manual switchboards. A connection is switched from the caller to the sender by selecting idle circuits from a pool and interconnecting them end-to-end. The connection is stable, exclusive to the parties, and quality is so good that an intercontinental call can hardly be distinguished from a local one.

The other network is an upstart. Having its roots in the U.S. Defense Department's Advanced Research Projects Agency network (ARPANET), the Internet began in 1969 as an experiment in packet-switched data networking. Today's Internet has evolved from its beginnings as the development of a coalition of government agencies, universities, and contractors. Several events launched the Internet as we know it today. First was the decision to release the Internet and its protocols into the public domain. The backbone was converted into a profit-making enterprise and Internet service providers (ISPs) emerged to manage and bill for customer access to the backbone. The second event was the development of hypertext transfer protocol (HTTP), which enabled the use of Web browsers. These relieved users of the need to understand arcane Unix commands and protocols, such as file transfer protocol (FTP) and telnet, which permits remote users to log on to a distant computer over the network. Email client software made it easy to attach data, voice, and video clips to email messages and launch them from the desktop. Domain registration processes were set in place to enable users worldwide to obtain unique addresses, or uniform resource locators (URLs).

Parallel to Internet development was the 802 initiative by the Institute of Electrical and Electronic Engineers (IEEE), which launched a project to standardize local area network (LAN) protocols. LANs had limited interplay with the Internet at first. Through the middle 1980s the standard desktop device for Internet access was a dumb terminal. As the LAN became popular, an explosion in the number of personal computers and the capability of networking them together enabled users sitting at their desks to obtain information from millions of computers throughout the world.

It's hard to realize that in just a few short years, the Internet has burgeoned into an open worldwide network that has become nearly as mission critical as the PSTN for many corporate entities and some residences. The protocol that makes the Internet possible is Internet protocol (IP). Whereas the switches that link the PSTN are proprietary in every respect except their line and trunk interfaces, IP is an open protocol that is evolving to carry telephone traffic through the use of associated protocols. Lines in a telephone system are the connections to the subscribers and trunks are the connections between switches.

Many people in the industry believe that voice and data will converge into a single packet-based transport network that carries both. The industry generally refers to this merging of networks as convergence. In addition, the terms IP telephony and voice over Internet protocol (VOIP) are also used interchangeably by many in the industry. For the sake of consistency, we will adopt the following terminology from the March 9, 2001, report of the Secretary General of the International Telecommunications Union (ITU) World Telecommunication Policy Forum (http://www.itu.int):

IP telephony is used as a generic term for the conveyance of voice, fax, and related services, partially or wholly over packet-switched IP-based networks. IP Telephony may also include applications that integrate/embed the transmission of voice and fax with other media such as text and images. In this report, the term IP telephony is used interchangeably with VOIP. A third term, internet telephony, is also used in the report when referring to IP telephony or VOIP conveyed partially or wholly over the Internet.

Video is not specifically mentioned, but it is included in our definition of IP telephony. Video transmission is minuscule in both voice and data networks today, but as packet networks evolve, it will become a more commonplace application.

While total convergence may eventually occur, the network that supports it will not be the Internet as we know it today. VOIP has many attractive advantages that the PSTN, with its limited channelized bandwidth, cannot support. The PSTN does, however, set up telephone calls quickly and easily. The quality of the PSTN is difficult for packet networks to duplicate and impossible for the Internet as it exists today. The very things that make the Internet so attractive render it unsuitable as a PSTN replacement. Its lack of centralized and proprietary control and discipline makes it difficult to keep latency, or packet delay, within the required limits. The pricing structure, which is not sensitive to either usage or distance, is appealing, but few ISPs are stable enough to be entrusted with an application as critical as the public telephone network.

This does not mean that packet networks can't carry voice as effectively as the PSTN, but the Internet is not the appropriate packet network. For IP telephony to be a serious competitor with the PSTN, a reliable packet network with built-in quality of service (QOS) mechanisms must emerge. Furthermore, with today's state-of-the art technology, the gap is wide between what is possible and what is actualized, what is open and what is proprietary, and what is based on mature standards as opposed to proposed or draft standards.

Standards

A major argument in favor of VOIP is the fact that its protocols are based on open standards, whereas the PSTN is closed except at the line and trunk interfaces. A major factor that has fueled the Internet is the ease with which the multitude of carriers can connect to the network with the assurance of interoperability. The idea of open protocols has been around for many years. Data communications got its start with proprietary protocols, of which IBM's systems network architecture (SNA) is a prime example. While proprietary protocols are good for the developer, particularly if it has market power, they limit the ability of the user to choose other products.

The International Standards Organization's open systems interconnect (OSI) model is a major attempt to standardize data communication protocols. OSI is not an architecture or a working protocol, but rather a seven-layer model on which other standards can be structured. Each of the layers is defined and serves as the basis for such protocols as IEEE's 802 LAN and X.25, which is the ITU standard interface to a packet network.

While the world was focused on the OSI model, the ARPANET project was taking a different tack in developing a fast and simple protocol, which became TCP/IP. TCP/IP is also a layered protocol that is similar in structure to OSI but was developed without the lengthy deliberations that characterize ITU initiatives. Although the Internet Engineering Task Force (IETF) did not set out to develop the de facto standard protocol for data communications, that has been the result. The success and versatility of IP have led many developers to focus on new applications including voice, video, and fax, which generally can be summarized as multimedia communications.

IP has become the de facto standard for data transmission, not necessarily because it is the best possible protocol, but because it has the support of all manufacturers and a tremendous amount of momentum. It is cheap and easy to implement, and it works. Practically every corporate data network runs over IP while the voice side of the business runs on the PSTN or a private network.

Transmission Media

The PSTN and IP have always shared transmission media. The PSTN runs over a digital backbone consisting of time-division multiplex (TDM) devices operating with fixed bandwidths of 64 kilobits per second (Kb/s). Until recently, analog circuits were the only media available for data transmission. When a consent decree broke up the Bell System in 1984, the PSTN was largely analog, and 9,600 bits per second (b/s) was about the fastest data rate the network would support. Data transmission lived uneasily in an analog telephone world running protocols that compensated for limited bandwidth and a high error rate.

In the mid-1980s, fiber optics was just beginning to come into play as interexchange carriers (IXCs) built their own networks to compete with AT&T. In less than a decade all that remained of the analog PSTN were the copper cable pairs that comprised the loop from the local central office to the subscriber's premises and a few remaining #1ESS electronic switching systems with analog circuit interfaces. An abundance of high-quality digital bandwidth soon replaced the restrictions the analog network had imposed. Dramatic increases in fiber and multiplexing equipment quality came on the market just in time to fuel the insatiable bandwidth demands of the Internet. In short order, the default bandwidth for data transmission vaulted from a 64 Kb/s circuit to full T1/E1 at 1.5/2.0 megabits per second (Mb/s). Today, bandwidths of as much as 10 gigabits per second (Gb/s) are available, and the limits have not been reached.

New Service Offerings

The abundance of high-quality bandwidth led to important new service offerings. Frame relay has been the most popular of these. Whereas other protocols such as X.25 packet switching take elaborate precautions to detect and correct errors, frame relay, relying on high-quality fiber-optic circuits, leaves it to higher protocols to arrange for retransmission as needed. The carriers that provide facilities for the Internet backbone offer IP services that are suitable for virtual private networks (VPNs). A VPN is an IP network that runs over the carrier's backbone. It has many of the characteristics of the public Internet but with a QOS that may be high enough to support voice and video.

Another key service offering is asynchronous transfer mode (ATM). ATM has the distinction of being the only protocol designed from the ground up to provide QOS for both voice and data. Any type of digital signal can be assigned to one of four service classes that range in quality from best-effort delivery to service with guaranteed stability. Data is sliced into short cells consisting of a 5-byte header and 48 bytes of payload and switched to the destination over a defined path. For several years, ATM was expected to become the one standard protocol for multimedia transport, but it has proven too expensive in comparison to more lightweight protocols such as Ethernet and IP.

Motivations for VOIP

Voice over IP has generated a lot of enthusiasm, but, like many new technologies, once reality sets in the industry realizes the expectations were too extravagant. The primary VOIP impetus came from equipment manufacturers who saw an opportunity for a greatly expanded market and from information technology managers who understood IP well but lacked an understanding of traditional voice networks. The arguments in favor of an all-IP network are appealing. If a data network is already in place, voice can ride practically free, particularly if the network has excess capacity. Even if voice isn't carried for nothing, IP promises to bring all the media under a common management platform, eliminating the need for separate voice, video, and data management.

Both of these arguments for VOIP have diminished as reality has set in. Voice does not ride free on a data network. More powerful routers, added costs of voice interface modules, and additional bandwidth add to the costs. The argument that the bandwidth is there anyway is valid only in building and campus networks. In the metropolitan and wide area network, if surplus bandwidth is available, new demands will come along to consume it, and the cost of adding more is often high. Another factor that nullifies the cost-saving argument is the fact that long distance costs have dropped so low that the payback period for the VOIP equipment required to avoid toll charges is usually too long to be attractive. VOIP is likely to pay in toll cost avoidance only if the company has a significant amount of international traffic and enough data bandwidth to support voice.

The second motivation, a common management platform, is more illusory than real. To be sure, the networks can be consolidated into a single data network, but introducing voice into the IP network complicates things. New protocols must be learned; gateway and gatekeeper functions, which are discussed later, must be managed; and router setup is complicated by the need to prioritize time-sensitive traffic. Managers are not insulated from the need to learn voice traffic management because elements such as gateways are potential choke points in the VOIP network.

Some vendors also suggest other reasons for VOIP. For example, the IP telephone can plug directly into an Ethernet wall jack and the desktop computer plugs into the telephone, so both can share the same wire run. In some isolated instances, this is beneficial, but the conditions are rare in which shared wire saves much. Offices that have been wired to the EIA/TIA 568 standards (http://www.tiaonline.org/) already have two or more wire runs. If the office is being newly wired, little is saved by reducing the second wire because the incremental labor and cost of the second wire run is much lower than the first. Wire congestion at the equipment room is, however, reduced.

The real value of VOIP lies in some interesting applications that are difficult or impossible with traditional networks. This book will discuss many of these. VOIP can be expected to be an important force in telecommunications technology, but it will not replace the PSTN in the foreseeable future; not perhaps ever. Users should not hesitate to purchase private branch exchanges (PBXs) for fear that they will soon be made obsolete by VOIP or the IP PBX. To the contrary, nearly all PBX manufacturers are adapting their products to support VOIP in manners that we will discuss.

To succeed in a totally converged network, VOIP must emulate the PSTN. Note that replicating the PSTN is not the issue. VOIP will be successful only if it provides enough increased value to make it worth the cost and effort for carriers to change their networks. To be attractive, the value must tilt in favor of VOIP as a result of cost reductions or service enhancements. The cost-reduction side of the equation is the least likely to prove successful. Although VOIP makes more efficient use of bandwidth, it comes about only through complexities that only a narrow spectrum of users and service providers will find attractive and only because of the cost-reduction potential. The service-side equation is where the industry needs to focus because VOIP enables services that are difficult or impossible to achieve on the PSTN.