IP-Based Voice Platforms Need a Common Standard

The transition to IP-based telecommunications within and
outside of the cable industry is looking less coherent than ever as individual strategic
requirements overwhelm any expectations that the dominant switched-circuit domain might be
quickly marginalized.

Over the past year, visionary leaders of companies like MCI
WorldCom's UUNet and Level 3 Communications Inc. have made a strong case for the economic
imperatives that are driving the Internet-protocol platform to replace the
switched-circuit infrastructure.

In fact, in many instances, they have suggested that the
costs of local calling, as well as long-distance calling, will be cut to practically

But while the case looks good on paper and on slide charts
at voice-over-IP conferences, the realities of short-term survival in the
IP-telecommunications world are undermining the chances for agreement on the universal
platform that would be essential to rendering the circuit-switched environment obsolete.

Nobody better illustrates the state of affairs than one of
the leading IP visionaries, Level 3 president and CEO James Crowe. He said his company now
has a first-line IP-voice system under beta trial, and he expects to launch services in
the third quarter, which would mark the first instance of direct competition with local,
as well as long-distance, switched service via the IP platform.

This means everything will be IP-based within Level 3's
operating "cloud," putting the carrier in position to work through alliances
with local service providers to deliver an end-to-end service in direct competition with
the incumbent local and long-distance carriers.

The architecture that Level 3 is deploying may or may not
win out in the standards battles, Crowe added, but the company stands to lose more by
waiting to get started than it does by having to make adjustments to be
standards-compliant later on.

"IP voice works today," Crowe said. "It's
not only less expensive, but we have an opportunity to see prices drop faster than we've
ever seen."

The cost of transport for one bit in voice transmissions is
14 times that of one bit in data transmissions, which means that voice is dominating the
economics of networking even as data surpasses it in total traffic volume, Crowe said.

"Unless we can get at that voice market, we're going
to be beholden to those who own the networks today in our efforts to move
information," he noted.

And there's the rub: As long as everybody isn't on the same
IP platform Level 3 uses, that carrier remains dependent on the existing switched-circuit
network to hand off calls for delivery to recipients on switched-circuit lines and to
intermediate in the hand-off between IP calls on Level 3's platform and the IP clouds of
other service providers.

The universal dependence on the existing
intelligent-network system to handle all of the complex call-setup and
feature-provisioning requirements between dissimilar IP clouds is a major barrier to
exploiting not only the cost benefits, but also the advanced features and operational
benefits of the IP domain, Crowe said. "That's why it's necessary that we agree on
some kind of standard and push forward," he added.

AT&T Corp. is in complete agreement with this,
according to chief technology officer David Nagel, who also serves as president of
AT&T Laboratories. The only problem is that AT&T doesn't concur with Level 3 on
what that universal platform should look like.

There's no doubting the cost imperative of the IP platform,
Nagel said, noting that moving a DS0 (64-kilobit-per-second) call into an OC-48
(2.5-gigabit-per-second) feed costs around $200 per port on a circuit switch, versus less
than $3 per port on a router.

This is why AT&T is intent on making fixed wireless at
300 kbps to 400 kbps work in its drive to provide first-line IP services nationwide so
that it can be all-IP-based in areas where it doesn't have cable connections, he added.

But even so, he said, AT&T and other carriers using
IP-voice systems are going to be dependent on the circuit-switched systems for a long time
to come. "The circuit networks aren't going to go away," he added. "The
question is how to interface with them."

Despite the widespread touting of the current H.323
standard as a solution to these issues, there really isn't a workable standard now, Nagel
said. "H.323 doesn't equal H.323 in different vendor implementations, and the OSS
[operations-support systems] are even more out of synch," he added.

While an initiative known as "iNOW" -- begun by
carrier ITXC Corp. and vendor VocalTec Communications Ltd. -- has expanded to include 18
other vendors in the effort to define a common H.323 profile, this version of H.323 does
not achieve the carrier-class level of support for IP telephony that AT&T and the rest
of the major players are looking for.

As a result, a "version 3" of H.323 is now under
development within the International Telecommunications Union. Version 3 really consists
of a variety of separate protocol initiatives, with the likelihood that the resulting
compilation of innovations will be even more diffused than version 2 has been.

Differing carrier agendas are making it hard to achieve
consensus on key issues in version 3, noted James Toga, director of systems architecture
for Metatel and a leader in the H.323 process.

In fact, he added, the goal isn't so much a single-platform
solution as it is meeting all of the requirements that are driving various initiatives
within each of the separate protocol pieces that comprise H.323.

"There are a bunch of different models to IP telephony
that lead to modifying H.323 in different ways, so there's a lot of push and pull in
different directions as we move to the next level," he said.

One of the most fundamental differences concerns the battle
between new-generation and incumbent carriers, he noted. "Is H.323 simply another
on-ramp to the PSTN [public switched telephone network], or is the PSTN over the long term
an on-ramp to IP telephony?" he asked. "The architecture you choose has a lot to
do with where your asset value resides."

In an area of keen interest to those on both sides of that
question, the H.323 group is a long way from resolving the issue over how the gateway is
modularized, noted Dale Scram, vice president of engineering at Ascend Communications Inc.
and a leader on this issue within the ITU standards-setting process.

The protocol to be used in managing hundreds of gateways
used for various applications -- such as multimedia access, call control and interfacing
with SS7 -- known as "H.GCP" (Gateway Control Protocol) is an "extremely
controversial issue, which I don't think will sort itself out until sometime next
year," Scram said.

H.323 is flawed in AT&T's eyes in part because there's
an absence of coordination between the gatekeeper and resource management, said Chuck
Kalmanek, an engineering executive at AT&T Labs.

Moreover, he added, the central-processing-unit-intensive
complexity of the H.323 gatekeeper makes it hard to evolve as new technical options emerge
and hard to scale as the messaging volume begins to overload the processing power of the

"We'd love to take advantage of an existing,
standards-based, off-the-shelf system, but we don't think that's the best approach,"
Kalmanek said.

He also took issue with the architecture preferred by Level
3 that's based on a call-control protocol known as MGCP (Multimedia Gateway Control
Protocol), which has been adapted for cable's use as part of the draft specifications to
be issued this summer by the PacketCable group at Cable Television Laboratories Inc.

AT&T's concerns with the MGCP approach have to do with
issues of scalability, security and flexibility in how service features are provisioned,
he said.

"Our goal is to balance all of the resources within
the broadband network so that we can provision the fullest range of services with maximum
efficiency," he noted.

As previously reported, AT&T is addressing these
concerns with the development of an alternative architecture known as "DOSA"
(Distributed Open-Signaling Architecture), which will also be included as part of the
specs in the phase-one segment of PacketCable's effort.

What all of the confusion adds up to is the fact that it's
everybody for themselves in the IP domain, with the PSTN system serving as the
life-support system for interconnection and provisioning of basic IN functions.

Even in cable and even within AT&T, there are likely to
be different approaches to the underlying architecture inside the IP clouds, as everyone
continues to feed off the PSTN lifeline, making it all but certain that the benefits will
be incremental, rather than revolutionary, for a long time to come.

"Once all of the switching is done at the IP layer,
then [delivering voice via IP] will be cheaper than using [class 5 switches],"
AT&T Broadband & Internet Services CTO Tony Werner said. "But in the
meantime, you have to get to a switch at some point in the network, so it's going to be a
gradual period of cost improvement until you can take full advantage of the