February 1999

IVR Makes Internet Telephony Sound Even Better

BY STEVE BAECHLE

Having only one ability - the potential to generate huge savings, for example - is no longer enough to make it big in the information technology industry. Like the young pitching phenomenon who can't bat or field, technologies such as Internet telephony only become viable if they advance in the fundamentals: namely, voice quality and network reliability. The solutions will get better and better, and steady progress has already left Internet telephony suitably rounded. So what's the new challenge? Features. It's not enough simply to configure high-quality voice. Carriers need an infrastructure that supports their actual services and replicates the voice world, with as little relearning and revamping as possible. But which features should come first?

The answer is simple: the ones associated with the early applications. To date, fax services, wholesalers, and pre- and post-paid calling card services have made the most of IP telephony. But deploying these services has brought some challenges bobbing to the surface. One major issue is that of interactive voice response (IVR), a key aspect in controlling overhead and administrating international card services.

THE HIDDEN COSTS AND RISKS OF CENTRALIZED IVR
In providing international calling card services, most carriers will not establish a physical presence in every country. They will maintain nodes in the most logical areas, and then lease bandwidth from other carriers in off-net countries. This is similar to the regular "voice world."

However, with IP telephony today, all of the billing, caller identification, and user prompting associated with calling card services have to be performed centrally. This is so primarily because gateways lack the built-in features and intelligence such as IVR and billing capabilities that are common to Central Office circuit-switches.

This centralized architecture means that every international calling card transaction must pass through the central database at least three times as calls progress from "please enter your calling card number and PIN," to "please enter the number you would like to call," and finally the placement of the call. Each user prompt along the way must come from the centralized IVR system, and each response must be sent back.

IP telephony is all about savings, but the overhead costs produced by these round trips across the network can put a real dent in ROI projections. If they're leasing capacity, carriers are paying other providers for bandwidth consumed by prompts as well as actual voice traffic. And because the prompts clog up the network, more bandwidth is required to ensure the highest quality voice. In an industry where profits and losses fluctuate in increments of megabits-per-second, the slightest waste can accumulate into huge sums.

But there is a way to avoid this problem. As we might expect, it includes duplicating the functionality and features of the voice world in packet-based IP networks.

ON THE EDGE, GETTING SMARTER
In order to reduce the number of round trips necessary to prompt calling card holders through to their destinations, carriers can look to push the IVR capabilities and functions of the central switch out to remote nodes. By integrating IVR capability into the IP telephony gateway, the prompting for customer calling card numbers and other items can be handled locally, through a distributed architecture.

Using a simple command language, gateways can be configured to perform voice prompts, calling card verification, and dial-digit processing at the remote sites so that only the actual long-distance voice traffic passes through the IP network as voice. Before, when the voice prompts such as "please reenter your card number now" were "spoken" by the central switch, they had to be sent across the network as actual voice traffic. With integrated IVR, the central switch sends IP messages to the local gateway at the remote nodes directing the IVR in the gateway to prompt the caller. Because the "speaking" of prompts takes place locally, the voice no longer has to be sent across the WAN; only the directives from the switch - relatively tiny IP packets - must go back and forth to the central database. This saves lots of bandwidth per call.

Pushing this capability out to remote gateways has multiple cost and performance benefits:

Lower Cost: The cost-savings generated by decentralizing IVR are multifaceted. First and foremost, it eliminates detours that send bandwidth costs rocketing. If a carrier is based in the United States, for example, and a cardholder places a call from Italy to Argentina, the call must detour to the United States three times at the carrier's expense. Each time, the carrier pays long-haulers to carry the traffic between its own nodes.

Before integrated IVR, carriers could only get around this by placing switches in each country, which as you can imagine, pushes costs through the roof. The integrated IVR capability allows these switching decisions to be made locally, thereby taking advantage of least cost routing (LCR).

Also, distributed IVR guards against a sort of fraud. With centralized IVR, the would-be caller uses bandwidth on the carrier's network before having their PIN validated. Distributed IVR ensures that precious bandwidth is only used by paying customers. Carriers also see costs reduced or avoided by virtue of not needing as much bandwidth in the first place, or not having to increase capacity as soon or as often.

There are some "softer" savings benefits as well. For one thing, fewer standalone components may be needed at each node, thus reducing capital investment. And, as a result of the added call-path flexibility, carriers can take advantage of local call routing plans and LCR schemes to reduce operating costs on a dynamic basis. Calls can be completed faster as well as less expensively.

Better Voice Quality: Being able to switch calls directly between remote locations without going through headquarters can improve voice quality as well as profits. Taking the shortest route between two nodes often reduces the number of hops and the amount of times the call has to be compressed and decompressed. This allows calls to be placed faster and preserves the integrity of each call.

Overall Network Reliability Increased: Integrating IVR into the IP telephony gateway is part of a sound overall reliability strategy. Unlike PC-based gateways and router solutions, a true carrier-grade gateway is engineered for reliability from the ground up. Such gateways could start with a parallel processing architecture that ensures complete redundancy and eliminates single points of failure within nodes.

Avoiding PC systems alleviates concerns about mean-time-between-failures (MTBF) on components such as disk drives, saving carriers from having to consider otherwise premature upgrades due to this concern. Hot-swapability is another must. Having to bring down the network, or even a segment or node, in order to add capacity or features means jeopardizing performance needlessly.

Like integrating IVR, integrating circuit-switching reinforces the seams between the packet- and circuit-switched domains. For real reliability, backup is needed. Integrating switching into the gateway makes this backup automatic, and also allows performance-driven control over service quality. For example, if the Internet gets congested or access fails, the gateway automatically routes traffic using the circuit-switched backup system.

In general, the strategy behind integrated IVR, circuit-switching, as well as providing an interface to billing systems and other features, is to get the functionality of as many different pieces as possible onto one fully redundant, highly scalable IP telephony platform.

IVR + IP = MVP
Though a rich selection of features is key to making Internet telephony viable for established voice providers, few will have as far reaching an effect as those that customers experience directly. In a sense, the IVR is the customer service representative during calling card transactions. This is not something providers can take lightly.

As Internet telephony infrastructures evolve, sophisticated gateways and distributed intelligence in the network will continue to make the business case better and more rock-solid.

While many heralded technologies end up dying on the vine, the awesome thing about Internet telephony is that the entire industry, even the players who seem to have the most to lose, have come forth in support of it. With such a public showing, and obvious end-user demand for savings and simplicity, we trust that integrated IVR is just one of many milestones on Internet telephony's path to the record books.

Steve Baechle is the vice president of technical consulting for Hypercom Network Systems. Hypercom provides IP telephony solutions to aggressive carriers and infrastructure equipment to enterprises seeking to benefit from convergence and integrated wide-area solutions. For more information, please visit their Web site at www.hypercom.com.