Pipe Dreams

IP already dominates computer-to-computer data communications, and although voice over IP (VoIP) is a small fraction of overall telecom traffic, it is growing as well. Forecasts predict the IP telephony market could reach $8 billion by 2003. But with unresolved standards issues and tremendous uncertainty in the market, IP telephony seems more like a pipe dream than it does a reality.

Many carriers aren't even willing to talk about IP telephony openly and even fewer have set deployment plans. Meanwhile, analysts predict wireless networks will experience the same growth in data traffic as wireline, albeit one or two years later. To date that growth has failed to materialize.

IP TELEPHONY TODAY Today, Internet telephony service providers (ITSPs) provide gateway servers in major metropolitan areas to interconnect public switched telephone networks (PSTNs) and IP networks. Subscribers connect to these gateways by dialing a local access number. The ITSP then carries voice communications in IP packets over the Internet to the terminating gateway at the destination, where a second local call is placed across the PSTN to complete the connection. This circumnavigates the long-distance interexchange telephone network.

ITSPs are providing a low-cost connection by using lower-cost general computers vs. purpose-built switching platforms. Until recently, this was a more favorable arbitrage environment because ITSPs were not subject to the access charges imposed on inter-exchange companies.

There are key areas for growth potential beyond today's ventures. For example, facsimile communications is less sensitive to the delays and latency problems often associated with shared packet networks such as the Internet. And for low-cost intercompany voice traffic, fewer features and lower quality is acceptable. For this traffic, VoIP gateways are deployed between the enterprise's PBX and router equipment to divert internal voice traffic onto the IP network.

For wireless carriers, VoIP is an alternative to the PSTN for routing backbone traffic. Earlier this year, WebTel Wireless first demonstrated the capability to make a quality VoIP call using PCS. According to Aaron Haskal, WebTel Wireless president & CEO, VoIP is a viable alternative to the PSTN.

"For cellular and PCS carriers that are dependent on interexchange carriers' high domestic and international charges, VoIP provides a way of bypassing the PSTN, resulting in a significant cost advantage," he said. "Voice quality appears to be equal or better than some international PSTN calls."

QUALITY & INTEROPERABILITY To ensure voice quality, IP traffic is transmitted over managed Internets. Today, managed Internet providers maintain acceptable voice quality by over-engineering network capacity to prevent lost packets and latency.

The industry is defining new IP-based standards to support different service levels, the prioritization of voice calls, IP switching and call control. Quality concerns over lost packets and latency are being addressed, and interfaces between PSTN and IP networks are being developed. Coupled with the deployment of managed Internets, solutions are within reach.

Lawrence Corvari, XYPoint vice president of engineering, likens VoIP quality issues to the voice quality of early TDMA systems. He said the industry will resolve these concerns within a year.

"For regional and small wireless carriers to compete, they need control of their backbone network -- VoIP technology provides this capability," Corvari added.

Interoperability is another concern that plagues VoIP gateways because different vendors will use various schemes for handling calls, facsimiles, voice encoding and traffic prioritization.

This year the ITU endorsed H.323 as the standard for inter-connectivity between IP equipment and the PSTN. The H.323 standard includes audio and video codecs, call and system control. Although major telecom and computer vendors support this standard, some companies are concerned about the lack of security, call prioritization, bandwidth-management standards, scaleability testing and message-set complexity. As a result, many vendors continue to pursue proprietary protocols.

Jerry Mistrot, telecom technologies director of custom development, plans to support both H.323 and SIP over proprietary system seven IP (SSIP), which maps SS7 to TCP/IP.

"H.323 requires a high volume of messages to set up a call, therefore as VoIP traffic volume ramps up, the complexity of the message set may exceed acceptable time constraints," he said. "We will support any and all call setup protocols while helping to migrate to the future protocols."

Although it is uncertain whether the market will accept alternative protocols, VoIP continues to escalate, with leading router suppliers entering the market and incorporating VoIP gateway functionality into router products.

Regardless of which VoIP standard emerges as the dominant solution, in the long term VoIP will not be the cheapest alternative. PSTN carriers will match any pricing advantage.

Therefore, the driving force for IP telephony is not short-term competition with the voice network, but rather the potential to deliver new services through the voice and data communications combination. This market is where wireless IP telephony will come to the forefront. The combinations will enable wireless access to the Web, extended virtual private networks to incorporate mobile users' unified mailbox and message management, and single personal numbering/addressing for all forms of communications as well as location and find-me services.

WIRELESS DATA A lack of nationwide coverage, high prices, limited bandwidth and end-user device limitations have hampered the wireless data market. But in 1999 several technology advances will address these shortcomings including increased data-access speeds and improved user interfaces such as microbrowsers.

"This is the first time that we are seeing essential technologies such as over-the-air packet data rates of 64kb/s to 144kb/s becoming available," said Steve Paolini, Nortel Networks director of wireless Internet marketing. "When this happens, carriers will be able to effectively market wireless data. The wireless data market is finally ready to take off. What is important now is the migration of the circuit-switched network to a packet-based network."

The GPRS specification was completed at the end of 1998, TDMA is adopting a modified version of this specification, and the cdma2000 1X specification is defined. Several vendors have announced that GSM, CDMA and TDMA networks will be able to provide packet-data rates up to 144kb/s by year-end. As a result, carriers will be able to provide packet-data services equivalent to PSTN and cable modems.

Specifications for 3G also are expected to be complete by year-end and promise to extend packet data up to 2Mb/s for low mobility indoor environments. In addition, a key component of the Wireless Application Protocol (WAP) Forum is to compress the capabilities of the specific air interface and mobile devices. This will enable a variety of mobile devices, each with potentially different display and capacity capabilities, to access Internet applications.

"The challenge is not to integrate wireless and wireline but to totally integrate all telecommunications encompassing wireless, wireline, the Internet and multimedia," said Lynn Whittington, Motorola Cellular Infrastructure chief technologist. "Operators will soon have the ability to uniformly offer services to their customers through any access device, subject only to the capabilities of that access device."

THE KILLER APP Wireless Internet is becoming a pivotal application for carriers and vendors. Experts expect the demand will follow the wireline model. Although access costs will be substantially higher than the wireline average of $20 per month for unlimited use, there are two key market aspects that have the potential to make wireless Internet extremely attractive.

Internet advertising revenues could help lower subscriber prices and make services more affordable. For example, in Hong Kong local restaurants advertise to subscribers in their vicinity using short message service. Carriers also can place a high premium on high-demand services such as wireless access to corporate databases.

"There is a need to make the enterprise IP data users go wireless, enabling people to be productive when away from their desk," said Dan MacDonald, Nokia Wireless Business Communications vice president of marketing. "Making the IP network mobile includes IP multimedia access across wireless LANs, including voice."

UNCERTAINTY It still is not clear whether standards or innovative proprietary approaches will become the dominant technologies in the future. Faced with such uncertainty and the entrepreneurial dynamics of the IP industry, vendors are building multiple alternatives.

John Shantz, Cisco vice president of market development, expects that initially there will be an IP overlay to the wireless-access networks.

"We must build solutions for both and look for opportunities to drive commonality in the standards," he said. "We must build products in such a way that the platforms converge."

Today, VoIP offers wireless carriers a viable and cost-effective alternative to the PSTN. Although VoIP is unlikely to sustain its price advantage in the long term, it is an important incremental step toward the transition to an IP network.

The ubiquity and unprecedented momentum of IP makes it a desirable long-term option.

To a limited extent, the migration to packet-based networks is happening today, as carriers replace proprietary equipment with client-server platforms to run applications. Several carriers already are migrating HLRs onto separate IN platforms. As data traffic increases, the cost threshold to migrate completely to a packet-based network will become evident.

If carriers continue to focus on spectrum capacity and network costs and fail to leverage IP technologies to their full potential, they could lose out on the opportunity to provide new services and gain new revenues.

In a packet network, you share resources, unlike a circuit-switched connection where you have reserved capacity for the duration of the call. ATM, which carriers are deploying to interconnect toll switches, also is a shared packet protocol. However, ATM provides different service levels and circuit emulation to sustain call quality.

Although standards exist for IP over ATM, the Internet Engineering Task Force is extending the TCP/IP protocol suite to support voice communications without the high overhead of the ATM protocol. However, there is no evidence that an IP network ultimately will be more cost-effective than an ATM network. The key advantage of IP over ATM is the ubiquitous use of IP and the potential for application development to proliferate.