IP-Phone-Service Rollouts Still Have Some Issues Left


San Jose, Calif.-The timing for commercial rollout of Internet-protocol telecommunications services over cable networks remains an open question as operators on the cutting edge of the deployment curve reported they're still waiting for key issues to be resolved.

"It's not that the technology doesn't work," Comcast Corp. vice president for strategic planning Steve Craddock said during a recent presentation at the Voice on the Net Conference here. But getting it to operate efficiently at scale is still a work in progress.

"In our humble opinion, what's required for primetime is increased network availability, low-power [consuming] clients [modems], DOCSIS [Data Over Cable Service Interface Specification] 1.1 [modems] and integrated support systems," Craddock said. "And if I want to do primary-line, I have to add to that a full feature set."

Most of the items on this short list are coming together more or less on schedule. But some issues-including the integrated-support-systems question-threaten to slow things down.

Another issue, having to do with integrating packet voice with other new technologies, stands out as a new, largely unanticipated conundrum in the ramp-up to IP telephony.

On the plus side, technical issues embodied in Cable Television Laboratories Inc.'s ongoing DOCSIS and PacketCable agendas are getting handled on or very close to anticipated schedules.

For example, CableLabs now expects to begin its certification process for modems conforming to DOCSIS version 1.1 in June, two months later than some had hoped.

Craddock said he expects to have DOCSIS 1.1 product to use by the fourth quarter, which will allow Comcast to move away from the "best-effort" restrictions of the 1.0 system it has been using in a test of IP telephony in Union City, N.J.

"With 1.1, we get dynamic QoS [quality of service]," Craddock said. "I can identify packet flows; I can assign priorities to flows; I can paint packets and do magical things with them."

These include the ability to implement a "very high degree of flexibility in managing the flows on the fly" and a "high degree of security," Craddock said, crediting AT & T Corp. for promoting the addition of "very good security hooks" to the protocols.

Comcast-which is currently evaluating the first phase of its IP-voice trial in Union City-expects to follow up with additional trials that will include testing of a second-line service, as well as a primary-line service, Craddock said.

"Probably about this time next year, we'll start to look at our first consumer deployments [of IP telephony]," he added.

Overall, the standards-compliance side of the cable industry's PacketCable initiative, which includes DOCSIS 1.1, is progressing well, noted Glenn Russell, project-development manager for PacketCable at CableLabs.

"Where we're at right now is really trying to grow and foster vendor participation in our CableLabs testing programs to accelerate interoperability and get products to market so that members can offer real services," Russell said.

"Our next steps are expansion of the PacketCable program, extending the architecture beyond [DOCSIS version] 1.0 to account for interdomain communications and definition of feature capabilities," he added.

Interdomain communications, or zone communications-in which various Packet-Cable-compliant service regions can interoperate with each other without having to hand signals off to the public switched telephone network-are vital to companies like Comcast, which has clustered Atlantic Seaboard franchises from northern Virginia to northern New Jersey to create a vast footprint for all-IP communications.

"By the first quarter [of 2001], we'll be looking at multizone, because basically, I want to get traffic off the PSTN and put it on IP backbones," Craddock said.

MSOs are also looking forward to introduction of the next generation of modems beyond DOCSIS 1.1, where the use of "Advanced Physical Layer," or "Advanced PHY," options will make it easier for them to meet the QoS requirements associated with delivery of commercial-grade telecommunications services.

"Advanced PHY allows us to run much higher constellation rates in the upstream, and it allows us to keep the stuff up without losing sessions," Craddock noted.

Comcast is preparing to launch a trial in Philadelphia that will employ PacketCable technology on a broader scale than the consumer telephony service being tested in Union City, Craddock said.

He added that the Philadelphia trial would use Cisco Systems Inc.'s platform, versus the Lucent Technologies and Motorola Corp. platform undergoing testing in Union City.

"We're really planning to have a lot of different types of services, IP telephony being one of them, but lots of integrated services flowing over this architecture," he said.

Where the prospects for moving to full-scale commercial launches of IP telephony and related services start to get murky is in systems management, where multiple layers of the cable operator's network and administrative systems have to be integrated into the packet-communications system.

"There's a new generation of OSS [operations-support systems] coming into play, and we have to have underlying support at the network-management layer to actually support all of those operations systems," said Mark Bakies, manager for cable communications at Cisco. "We have to let them actually look into an IP network, which is a fairly dynamic entity in and of itself, and be able to correlate across multiple network elements."

Bakies added that this means the system must instantly be able to provide answers to questions such as: "Where does the call go, and why did that call drop? Was it because the router reset, or something that happened on the PSTN side of the network?"

He continued: "If you start building from the bottom up, from the network management through the operations systems, there's an awful lot of work that needs to be done. Bringing the horsepower of 10 or 15 different companies together to start to get some of the next-generation operations-support systems together is where a lot of our focus is, along with taking those OSSs and bringing them back to the RF architecture."

Craddock backed that assessment, noting that a "lot of vendors have uniquely strong skills in one area, but you can be strong in one area and not in OSS. So we're trying to get them to put the hooks in so that we can take best-in-class and integrate it into services and systems we have today."

While PacketCable was designed to take advantage of existing telco operating systems-as Cisco is doing with Telcordia Technologies Inc. on Le Groupe Vidéotron Ltée's commercial IP-voice service in Canada-the truth is that the "operating systems of the PSTN don't really work very well" in the packet-voice environment, Bakies said.

In February-two months later than expected-Vidéotron moved to live connections of nonpaying customers in its ramp-up to commercial rollout. Now, the Canadian MSO continues to work through this phase of testing prior to launching a market trial to several thousand households, which executives had said would begin in the first quarter.

Still another issue looms on the IP-voice horizon that operators and CableLabs must sort through if they are to achieve all of the advanced service capabilities they aspire to.

This involves the ability to accommodate the addition of new networking and appliance components at the customer premises without going beyond the latency limits set for voice service.

"What happens when [the IP-voice stream from the home-interface unit] doesn't hit a solid telephone line, but gets plugged into the home LAN [local-area network] on an 802.11B [Ethernet jack]?" Craddock asked, adding that things get even more latency-sensitive when the subscriber adds a cordless phone to the mix.

"Somebody has to be thinking about all of this additional latency," he said.