PacketCable Advances, But IP Not Close


The PacketCable group is close to issuing its first set of
specifications, but it seems a long way from convincing many skeptics in the engineering
community that IP telephony will be a viable option within the next year-and-a-half or so.

Senior engineers with some of the key MSO backers of the
PacketCable initiative suggested that true first-line-quality service over
Internet-protocol channels might take much longer to achieve, despite considerable
progress on several fronts within the vendor and operating spheres outside of cable.

For example, Level 3 Communications Inc. and Bell
Communications Research (Bellcore) have agreed on a key IP-telephony protocol under
consideration by the Internet Engineering Task Force that dovetails nicely with the
direction that PacketCable is going in, they said. But, engineers added, this doesn't
mean that the product base for interoperable systems under that protocol will materialize
anytime soon.

Mark Coblitz, vice president of strategic planning for
Comcast Corp. and chairman of the PacketCable task force, has made it clear that he
doesn't view IP telephony as a near-term option for cable. But he and his colleagues,
working directly with Cable Television Laboratories Inc. and the vendor community, have
voiced confidence that the technology would offer a viable first-line-service option for
the cable industry by sometime in 2000.

Many representatives from the vendor community have
stressed their confidence in their ability to meet this target, as well. Cisco Systems
Inc. and Bellcore -- as suppliers to both Le Groupe Vidéotron Itée and Sprint Corp. in
the latter's ION (Integrated On- Demand Network) voice-over-IP initiative -- said
they intend to be supplying first-line systems by the second half of next year.

But not everyone believes that this translates into a quick
leap to a cable-optimized option.

"There are a lot of hoops that you have to jump
through to overcome the latency and other issues," said James Chiddix, senior vice
president of engineering for Time Warner Cable. "IP is fundamentally a bad transport
structure for video, voice or any isochronous communication."

Time Warner is a founding member in the PacketCable
initiative, along with Tele-Communications Inc., Comcast, Cox Communications Inc.,
MediaOne Group, Rogers Cablesystems Inc., Cogeco Cable Canada Inc. and Vidéotron.

If and when Time Warner moves to offering voice services
beyond its initial service base in Rochester, N.Y., such a step won't be a function
of the availability of the IP-voice option, Chiddix said, echoing widely held sentiments
in cable.

"Holding back on taking advantage of a business
opportunity so that you can take advantage of a new technology is usually a disappointing
process," he added.

Asked his opinion on the ability of Vidéotron to deliver
on its recently announced commitment to getting IP-voice services under way commercially
before the end of 1999, Chiddix expressed doubts that the Canadian MSO would be able to
provide wide-scale toll-quality service within that time frame.

TCI and its prospective new owner, AT&T Corp., have
heavily touted the integrated-service potential of IP voice mixed with other forms of IP
data over broadband links. The companies, however, have focused their plans for
telephony-service launches over TCI networks in the year ahead on the use of proprietary
systems, starting with the Arris Interactive-supplied cable-phone technology currently
used by TCI in Hartford, Conn.

When asked whether TCI would be moving to the IP platform
within the 18-month time frame mapped by the PacketCable group, a senior executive at the
MSO rolled his eyes and shook his head.

"No comment," the executive said.

On the Level 3-Bellcore compromise, "You can say that
this is progress," said one engineering executive, asking not to be named. "But
when you consider that the IETF is having to go in and rip out some of the existing
[H.323] standard to accommodate the new one, you're forced to recognize that things
are moving slower than some of the optimists would have had us believe a few months

Level 3 is a next-generation IP carrier that has said that
it will be in a position to begin offering first-line services on a wide scale next year.
It and Bellcore recently reached agreement on a specification for delivering SS7
(Signaling System 7) and other intelligent-network functions -- special services like
three-way calling -- over the IP network.

Level 3 was proposing a protocol dubbed "IDCP"
(Internet-device-control protocol), as opposed to the SGCP (single-gateway-control
protocol) proposed by Bellcore and Cisco.

PacketCable was already moving toward endorsing the
Bellcore/Cisco solution, and it can now expect to see wider vendor support for its choice,
officials said. They noted that the merged specification, known as MGCP (media GCP), is to
be made available without a fee to service providers and other vendors.

The integration of the call-control mechanisms of the PSTN
(public switched telephone network) and IP-telephony networks will enable customers to
access services, including voice and fax, without modifying existing telephone and fax
equipment or dialing access codes, said Christian Huitema, chief scientist for Internet
architecture at Bellcore.

"The industry was faced with either picking one
protocol or implementing both," Huitema noted. "By merging the two proposals, we
resolve the dilemma, provide a safer environment for the manufacture of telephony gateways
and ensure the development of the call-agent architecture."

The Level 3/Bellcore pact was an important step in the
right direction. But it was only one of many that must be taken before cable can expect
all participating networks to be linked via the same call-management, network-operations,
billing and other protocols, noted Scott St. Clair, director of corporate communications
for 8x8 Inc., a member of the PacketCable royalty-free licensing pool.

While the company expects testing of PacketCable-compliant
systems to start by the end of next year, it doesn't anticipate wide-scale commercial
deployments until late 2000 or 2001, St. Clair said.

"Certainly, the perception among MSOs is not that
they're going to be able to wrest the local-telephony business from the Baby Bells
and GTE [Corp.] by offering calls at two cents less per minute," St. Clair said.
"It has to be an integrated package of services that offers something different to

Just what that package might look like can be seen in the
specifications taking shape within the PacketCable group. The first set of specs, targeted
for issuance in December, will focus on a consumer-oriented first-line service, with
business-level and video-telephony specs to follow in 1999.

The consumer service will use MTAs -- multimedia terminal
adapters, pegged to architecture proposed by 8x8 and others -- to pass signals back and
forth between the cable-data network and standard Touch-Tone telephone sets, serving up to
four separate phone links per household. These MTAs can be integrated components of
stand-alone DOCSIS (Data Over Cable Service/Interoperability Specification) modems or
"whole-house" line-management devices that connect to all of the phones in the

According to the latest thinking within the PacketCable
group, this "Phase 1" version of the PacketCable product line must support:

• CLASS (customer local-area-signal service) and other
enhanced-calling features;

• Toll-quality voice and network delays of less than
200 milliseconds end-to-end in one direction;

• Clear channel transmission for Touch-Tone signals;

• Backup powering via cable line or batteries;

• Emergency and operator services;

• Real-time fax;

• Standard analog-modem connections;

• The E.164 standard-numbering plan (where each cable
customer's packet address is given a number to accommodate connections from the
outside PSTN); and

• Both three-way calling using the embedded client and
four-way or more party conferencing in conjunction with the use of a centrally managed
"multipoint controller."

Each local-calling-area cluster within the PacketCable
"cloud" must be interconnected with other clusters through standardized
interfaces that ensure seamless operations for what are termed "on-net" calls,
which are calls from one PacketCable customer to another, no matter where they are located
around the country.

This means that participating companies must agree on means
of handling back-office administration such as billing, provisioning, network monitoring
and management, as well as passing signals through gateways to the outer PSTN network or
to adjacent PacketCable networks.

All of this represents a very tall order when it comes to
setting specs prior to accumulating considerable field experience, noted Don Burt,
president of Probita Inc., an engineering-consulting and systems-integration-engineering
company that is working on DOCSIS, OpenCable set-top and PacketCable issues. "Getting
fully interoperable, scalable systems is going to take longer than many people
realize," Burt said.

In fact, Burt added, deriving the whole set of specs before
going to early commercial deployments could be a mistake, given the surprises that could
lurk in the field experiences to come and the freedom that vendors will need to adjust
their systems to accommodate those issues.

"Since a lot of these issues can be mediated in
software, it might make sense to allow more flexibility early on to accomplish
interoperability quicker in the long run," Burt said.