BY JONATHAN ROSENBERG

Anyone involved in the Voice over IP (VoIP) industry over the last few years has seen that there have been many twists and turns on the road to deployment. New protocols, such as MGCP (Media Gateway Control Protocol) and SIP (Session Initiation Protocol), have entered the arena and caused changes in directions. New architectures, such as the softswitch, have challenged service providers to think carefully about how interfaces to the traditional SS7 networks will be handled. Many industry observers believe another major shift lies ahead: the emergence of true converged applications that will integrate the "brains and brawn" -- the intelligent architecture and carrier-grade versatility and scalability -- that service providers of all sizes will need.

Service providers and vendors have certainly talked about services and features, but so far these have been slow in coming. The primary applications of VoIP to date have been toll bypass and PC-to-phone calling -- all with the end goal of "cheap long distance." But the price differential of VoIP may be short-lived.

One need only watch the media to see the constant barrage of commercials advertising lower and lower prices for long distance calls. Within the United States, long distance calling has gone from distance-sensitive to distance-insensitive, and has dropped in cost from 20 cents to 10, to 7, and now as low as 5 cents a minute. Wireless providers are already experimenting with flat-rate pricing, whereby calls to a limited set of phone numbers are flat rate. A long-distance provider in Canada is already providing flat-rate service anywhere within the country. International calling has seen a similar decline in rates.

Given, then, that IP telephony is not likely to be much (if at all) cheaper than traditional telephony, and given that it will not have the same maturity and features users expect from traditional telephony for some time, why should they accept it?

The answer lies in the advent of true converged services. This means not simply porting traditional features like VPN and Centrex to IP. True converged features are about integrating Web, e-mail, presence, chat, and instant messaging with voice. This integration allows these other IP applications to become integral components of new voice services. These kinds of services, fundamentally, cannot be done in the circuit-switched world. They provide something new, something different -- something of value for service providers and their customers.

Converged services and applications also open the door to service provider differentiation. How? By taking advantage of what we may call "Rosenberg's Law." This theorem holds that the set of features and services that a service provider can deliver increases exponentially with the number of applications (such as Web, e-mail, and voice) that are combined to provide them. For example, take transfer, add Web, and you've got two service variants. Add e-mail, and you've got four. With such a large set of services, it's a fertile ground for differentiation.

What, specifically, will these converged services look like? Consider a few examples.

Web IVR
Interactive Voice Response (IVR) systems were supposed to make it easier for consumers to get the information they need from businesses. However, for most people, they are a nightmare to navigate and can often be frustrating. No two IVR systems use the same prompts or have the same structure. Consumers often have trouble remembering which option is the one they want, and have difficulty figuring out how to reach an actual human being.

Not surprisingly, the Web is a much more natural means for navigating these systems. Instead of pressing buttons on phones, users can click on links. Going back is as simple as hitting the back button on the browser. Users can quickly scan the available links at any level to find the right place for them.

The idea behind Web IVR is to combine the rapid communications of voice with the ease of navigation of the Web. Using their IP phones, users can call a number. During the day, they might be directly connected to a person. After business hours, they can be redirected to a Web page that contains a Web version of the IVR system. As users click through the system, they may eventually reach a point where a click actually completes the phone call to a real person.

Multimedia Caller ID
While caller ID is a popular consumer service, it still provides only limited information about the caller: phone number, and occasionally, name. The data is usually presented in a simple one- or two-line LCD display. What if, instead, a wide range of multimedia were be used to identify the caller? The multimedia could include text, audio, images, and even videos. We refer to this enhanced caller ID service as multimedia caller ID.

The most compelling example is a simple thumbnail photo of the caller. This would be ideal for people who are good at remembering faces, but not names. When someone calls, both his or her picture, along with the name, would appear. Another useful piece of information is a small audio snippet containing the caller's spoken name. Instead of just ringing, the phone might actually ring and then deliver an audio message such as "call from John Smith" in the actual voice of the caller.

Presence Notification Of Availability
The simple telephone has served as an ideal tool for real-time communications between people. But for the purposes of leaving messages, many users prefer e-mail, since it is more concise and is more easily stored and manipulated. For these users, a call that ends in voice mail is a useless call. To solve this problem, traditional buddy lists and presence systems can be combined with voice.

Buddy lists allow a user to subscribe to a list of friends. When one of those friends comes online, the user is notified, and they then have the option of sending an instant message. When coupled with voice, the service becomes very compelling. Rather than presence being a simple "on" or "off," indicating the ability to receive instant messages, it can reflect a person's readiness and ability to communicate using a variety of different means. For example, a user can subscribe to the cell phone of a second user. When the second user turns his or her cell phone on, the first user gets a notification, and can then send e-mail.

Another useful variation on this service is to allow one user to be notified of status changes on another user's traditional telephone. This service is used often in PBXs to allow an assistant to know when the boss is on or off the phone. With the Internet, this service can be made global. Even when the boss may be in another country, the assistant can subscribe to his or her phone to gauge availability.

WHY SIP?
The Session Initiation Protocol (SIP), specified in IETF RFC2543, is a powerful tool for call control and signaling that is gaining tremendous support among service providers and vendors. SIP turns out to be an ideal protocol for providing truly converged applications. This is primarily because it borrows so heavily from other Internet protocols, and in particular, HTTP and SMTP. Three features render SIP ideal for converged services:

1. MIME. MIME stands for Multipurpose Internet Mail Extensions. MIME is now used in a number of other protocols, including HTTP. It has become the de facto standard for describing all sorts of content on the Internet. Every audio format, every video type, and every image is a registered MIME type. SIP uses MIME, and as a result, SIP messages can contain Java applets, images (for multimedia caller ID), audio files (to announce the name of the caller), authorization tokens, or even billing data.
2. URLs (Universal Resource Locators). URLs have become the Internet standard way of addressing. Web pages are identified by HTTP URLs. E-mail is sent by clicking on "mail to" URLs. As new applications are developed, they define URL types as well. As a result, URLs have become the universal way to access Internet applications. Rather than defining some new type of address space, SIP uses URLs, and only URLs, for addressing. This means that it is just as easy to redirect someone to another phone as it is to redirect someone to a Web page. Transferring to a PC voice terminal is the same as transferring to email.
3. DNS. The Domain Name System (DNS) is the global directory service used on the Internet for translating domain names (such as www.yahoo.com) into IP addresses. The DNS is the key component needed to deliver Web and e-mail requests to the server that can appropriately handle them. Rather than inventing new mechanisms to accomplish this for VoIP calls, SIP borrows from the e-mail model. It uses DNS in exactly the same way e-mail routing makes use of DNS. The substantive benefit of this approach is that companies can share their e-mail routing infrastructure for voice call routing as well. This also simplifies integration of voice and e-mail. Servers along the call path can easily create and forward e-mail messages, and vice versa, enabling various combined services.

A PLATFORM FOR DELIVERING APPLICATIONS
Given that these converged services are the key to revenue for Internet telephony, what does a service provider need to deliver them? The answer: a converged applications server. This server is contacted by softswitches, PCs, and gateways when an application needs to be invoked for a call. In order to deliver converged services, such a platform must support not simply voice, but e-mail, Web, instant messaging, and presence. Its service creation tools must mirror those used on the Web, rather than those used in the IN (Intelligent Network) world. With such a server in place, the opportunities for service providers to deliver new services -- and the revenues associated with them -- are virtually limitless.

Jonathan Rosenberg is chief scientist for dynamicsoft. For more information, visit www.dynamicsoft.com. Also, Rosenberg co-chairs the Internet Engineering Task Force's (IETF) SIP working group, and chairs the IETF's IP Telephony working group.