Dallas -- The industry's technical leaders headed here in
droves last week for two days of saturation learning about trends in packet technologies,
software and network management.
The venue was the Society of Cable Telecommunications
Engineers' annual Conference on Emerging Technologies, which the SCTE's new president,
John Clark, called "the hidden gem of the industry."
Clark, who took over the helm at the SCTE last year, opened
the two-day conference at the Wyndham Anatole Hotel here last Wednesday, saying,
"Some call this the age of engineering empowerment, with new services, markets,
customers, revenues and, most important, profits."
He added that he intends to spend much of his time raising
corporate cable's awareness of the SCTE and its activities.
Attendance numbers were not available at deadline, but they
appeared to be tracking with traditional levels of 1,500 or so. The conference meeting
room was densely populated, as attendees listened intently to a steady stream of technical
The first of those came from Mark Dzuban, director of
technical-business development for AT&T Corp., who keynoted the conference. Dzuban
called cable's hybrid fiber-coaxial networks the best and most cost-effective method to
delivered integrated services.
"HFC is paramount to the vision for AT&T, and it
has great and major advantages over the alternatives," including xDSL (digital
subscriber line), fixed wireless and satellite techniques, Dzuban said.
Dzuban added that AT&T's current network contains six
to eight layers of transport and switching, and its new plan is to go with an integrated
approach that puts those layers on a common platform.
He said the development of standards-based cable modems was
mostly what drove AT&T to forge relationships with the cable industry, including its
planned acquisition of Tele-Communications Inc.
"[Cable modems] allow us to really leverage a common
platform ... so that we can move forward in a very rapid rollout plan," Dzuban said,
adding, "it's a quantum leap."
He said AT&T's network is evolving to a core high-speed
transport and switching platform, based on Internet protocol, and it will support
commercial and consumer applications.
But, he added, one element is still missing: a national
footprint that covers the 100 million households in the United States.
"We have HFC as a primary solution, but we need to
cover a national footprint," Dzuban said.
He declined to elaborate on AT&T's plans to round out
that gap, except to say, "There are solutions in other markets."
He said AT&T will enter the IP-telephony game "in
a big way" in 2000, noting that the company will use circuit-switched techniques, and
not IP, in 10 cable-telephony trials this year.
When asked if cable plant can adequately support the
multiplicity of advanced services that AT&T intends to provide, Dzuban was optimistic.
"Yes," he said, "I believe [that HFC plant] can meet the requirements that
He went on to say that cable networks "deliver -- when
operated properly, they do very well."
Asked to elaborate on the effects of proper operation on
field personnel, Dzuban said he was referring to mean-time-to-repair statistics that will
dictate telephony applications.
"We have to do that with reasonable consistency,"
Other panelists put a heavy emphasis on packet
technologies, saying that the move from analog RF to digital and IP will mark a radical
change in cable technologies.
On the packet front, Terry Shaw, project director of
network systems for Cable Television Laboratories Inc., detailed the PacketCable effort,
and executives with Cisco Systems Inc. and Lucent Technologies described how IP-telephony
services will come together on cable's HFC networks.
Harrison Miles, director of technical marketing for
Lucent's IP Communications unit, said he expects to see widespread IP-voice testing by
MSOs in the second half of this year, with some commercial rollouts as early as the first
quarter of 2000.
But he believes that it will take until 2004 before all of
the pieces are in place to support the full potential of IP telecommunications, weaving
together multimedia, voice and videoconferencing applications.
Miles emphasized that the performance parameters of the
primary IP applications -- voice, fax and multimedia -- are alike. He said latency time,
packet loss and jitter -- three ways to measure IP-data performance -- are similar for
cable-delivered packet services.
He also said he's not convinced that SGCP (Single Gateway
Control Protocol), which is currently endorsed by PacketCable, is the best choice for IP
communications, because it doesn't have built-in intelligence features that could make
IP-phone service as fully featured as regular telco phone service.
"When you go into the database [of the public switched
telephone network], you need to be able to look up and assess what the available resources
are. SGCP doesn't have all of the hooks to solve this issue," he added.
That belief may have an effect on PacketCable.
"Different entities are asking for different capabilities, depending on their legacy
situations," he said.
Miles said AT&T proposed DOSA (Distributed Operations
System Architecture) as an alternative to the SGCP approach. DOSA "envisions taking
the 5ESS switch and tearing it apart and distributing it in the network," he added.
DOSA works with SIP (Session Initiation Protocol), which
goes in the opposite architectural direction from the one taken by SGCP. SGCP relies on a
central call agent to communicate call-setup and other instructions to end-user points,
such as the packet-phone converter that will reside in cable modems interfacing with