IP telephony deployments and trials will remain sparse until operators
and vendors work out the technical and operational kinks

IP telephony's big payday has been somewhat of a moving target. Just when it looks like the technology and business of IP telephony are about to converge and finally cash in, a number of influential folks in the industry say it's still at least another year away.

That cycle remains in play this year, as most in the industry don't see VoIP to be ready for mass deployments until at least 2004. The less optimistic see IP telephony as a 2005 odyssey.

VoIP "is a lot like the pot of gold at the end of the rainbow. It always appears like it's just over the hill," says Joe Van Loan, senior vice president of technology for Mediacom Communications, one of several MSOs taking a closer look at how VoIP might fit into their future plans.

Comcast President Steve Burke put a scare into some manufacturers earlier this year when he declared that VoIP wouldn't be ready for prime time for another 12 to 18 months.

Despite what could be considered a slower than anticipated forecast for VoIP, Burke's comments were not about things like Dynamic Quality of Service (DQoS), CALEA or any other technical or regulatory wrinkle that needs ironing out. Instead, they were about when he believed IP telephony would be ready to make a sizable impact on Comcast's bottom line.

Putting those comments into further context, Sam Chernak, Comcast's vice president of VoIP, says those comments had little to do with VoIP: the technology, but VoIP: the business, and its status as "a material contributor" to Comcast's revenue plan.


But why does cable-based VoIP, at least its carrier-class, primary-line iteration, still appear to be in a holding pattern? Although IP telephony is certainly closer to becoming a solid, viable business than it was at this time last year, several obstacles remain, including system integration, scalability, and economics, to name but a few.

The integration of the PacketCable architecture (see full story, InDepth, p. 16) and how smoothly it hooks into the back office remains a key barrier that the industry must overcome.

"Each part [of the PacketCable infrastructure] might be mature in its own right, but the integration of those parts is relatively immature," says Jim Lakin, president of Arris Broadband.

How quickly VoIP can be made to scale is yet another wrinkle-filled problem. Although trials will help operators figure out how to offer the service to a few hundred customers, delivering it to 50,000 presents a quantum-leap challenge.

Scaling IP telephony "will be a big issue for tier-1 MSOs," says Apollo Guy, director of marketing and business development for OSS provider Lemur Networks. "The question is, how do you do this in a fully-loaded scenario...and make sure the service is equal to the incumbent's?"

Before that can happen, the platform itself must become much less complex, offers Dave Spear, executive vice president, strategy and market development for Cedar Point Communications, whose SAFARI C3 platform fuses several components of the PacketCable network into one device.

"I've believed from Day One that the only thing that's going to allow you to accelerate growth, in terms of VoIP in the marketplace, is that you have to bring a simple solution to the marketplace," says Spear.

Technical issues aside, some operators are unconvinced that carrier-class IP telephony provides the requisite financial benefits or that the technology is ready for widespread deployment.

In a recently released white paper, Cox Communications notes that, despite the attractiveness of IP telephony, the cost comparisons between the new platform and traditional circuit switched technology are "overstated."

Regardless of notions and predictions that VoIP costs will be less than half of those for circuit-switched technology, Cox notes that those figures typically exclude costs related to the network and other elements of the transport architecture. That puts VoIP cost improvements down toward the 8 percent to 10 percent range.

As the paper quoted Cox Senior Vice President of Engineering and Chief Technical Officer Chris Bowick: "There are numerous outstanding issues concerning equipment, powering options, integration of devices, etc., that make it imprudent to draw too many conclusions–especially with financial considerations."

In hard numbers, Cox writes that the cost per customer for traditional circuit-switched, primary-line phone service (with an average of 1.3 lines per customer) is roughly $610, assuming a 20 percent penetration of homes passed.

The "apples-to-apples" cost per customer for a primary-line softswitch solution (with network powering) is $564, Cox estimates, noting that the 8 percent cost savings partly comes from the lower cost of the existing cable modem termination system (CMTS) versus a Headend Interface Terminal (HIT) required for the circuit-switched deployments.

Per customer costs drop significantly– about 34 percent versus circuit-switched–to $404 per customer in a VoIP deployment based on home-based powering. Because of the different powering method, the comparison is not primary-line and, therefore, not apples-to-apples, Cox points out.

That's not to say, though, that Cox won't deploy VoIP when it believes the working parts have solidified and that a business plan can be written around it. In fact, Cox has a number of VoIP tests already in the pipeline.

But until proven otherwise, Cox said it won't launch IP telephony "until the technology is ready for widespread deployment and when it makes prudent business sense." Case-in-point: Cox opted for circuit-switched technology for its telephony launch in Wichita, Kan. earlier this year.

But what Cox presented and what's involved in a full-IP offering does not provide a balanced comparison, because voice is not the end-game of the PacketCable infrastructure, argues Maryling Yu, senior director of marketing with softswitch vendor Syndeo Corp.

"We have said all along what the Cox white paper says, which is we need to add up the circuit costs and the PacketCable costs," Yu says. "But the thing that they're forgetting, though, is the build-out for circuit switch voice is only applicable to all of those costs have to be carried by the voice business."

But offering VoIP via PacketCable allows operators to leverage an investment they've already made to offer high-speed data. "Now, you're going to overlay voice on it, which will then overlay multimedia, gaming sessions and video conferencing," she says.

While Cox continues to weigh its VoIP plans, other operators appear to be waiting to sift through the results of Comcast Corp.'s ambitious VoIP trial in the Philadelphia area (see case study, this page).

"We've talked to the top-20 MSOs, and what we've found is that a significant number of them are really waiting on Comcast," says Syndeo Co-Founder and Chairman Ted Griggs. "They want to see the bill of materials and [replicate] it. They don't want to be on the bleeding edge, but want to let Comcast work the rough edges out."

Still, other MSOs such as Armstrong Cable, Cablevision Systems Corp., Charter Communications, and Time Warner Cable have all dabbled or are about to start dabbling in the service to some degree.

Time Warner, for example, has balked at the high network powering costs associated with "lifeline" IP telephony so far, opting instead for a non-primary line service in cities such as Portland.

"This is not a one-size-fits-all marketplace," Lakin acknowledges. "[Operators] are looking at one or more service models, and rightly so."

Another obstacle that's hindering VoIP's rise is that the squeaky wheel tends to get the grease, which, for MSOs, translates into attention and cash. Most of cable's competitive pressure is coming on the video side of the house from the DBS providers, leaving VoIP developments somewhat on the backburner.

"The challenge, frankly, is competing with the other priorities they have as a cable company," says Brett Azuma, senior vice president of marketing with IP Unity, a maker of PacketCable-based media servers.


Despite the questionable short-term outlook for IP telephony, some developing trends will play a big part in cable's VoIP future.

While large operators like Comcast are taking on big VoIP projects largely on their own, a group of host-based companies such as Net2Phone Inc. and Gemini Voice Solutions are generating attention from MSOs that don't want to take on everything, including the capital outlay, all on their own.

For better or for worse, the hosted VoIP service model looks very similar to the way cable modem service got off the ground in the early days via partnerships with companies like Excite@Home, High Speed Access Corp. and ISP Channel.

Despite the fact those HSD "turn-key" providers went down the tubes for a number of reasons in and out of their control, the hosted VoIP segment is expected to grow into a $36.5 billion market by 2008, according to a recent Allied Business Intelligence forecast.

The hosted approach will likely attract operators that don't have all of the in-house expertise or the capital budget to offer IP telephony. Host-based firms can offer help in both departments.

Net2Phone, for example, handles the capital requirements for the VoIP system, except for the CMTSs and MTAs, explains Mike Pastor, vice president of Net2Phone's cable technology division, which is testing the approach in Puerto Rico with Liberty Cablevision (see case study, ID p. 6).

The typical range for a VoIP launch with premise-based powering can run between $400 to $500 per line. "We can make a third or half of that go away on the front-end," Pastor says.

A VoIP partner like Net2Phone makes its money via a recurring fee based on the upfront costs, which are spread throughout the life of the contract. Net2Phone also receives a portion of the subscriber revenue.

"It changes margins a bit, but [the operator's] bottom line is left the same or improved in the hosted model," Pastor says.

The model seems to be generating interest. "Even tier-1 operators that didn't pay attention to us 12 months ago are taking another look at this," he adds.


Instead of seeking VoIP fortunes via full PacketCable deployments or partnerships, some MSOs could throw a few scraps of meat to a starving segment of competitive local exchange carriers (CLECs). Instead of going the softswitch path, operators can deploy IP in the local loop and lease space on a CLEC-owned Class-5 switch.

Like the hosted method, the CLEC partnership approach can enable operators to offer IP telephony without breaking the bank on equipment and the people to run the service.

Plus, the CLECs have already met the operational and regulatory requirements, including number portability, billing, directory assistance, exchange agreements and adherence to emergency 911 and the Communications Assistance for Law Enforcement Act (see story, ID p. 9 for more on CALEA).

If an MSO tackled this from scratch, "we'd have to systematically build it or find companies and do deals with them," Mediacom's Van Loan says. "That would require a huge investment. We see [the CLEC partnership] as a way to start-up and build a business. It has some appeal to us."

Partnering with CLECs can also give operators a migration path to a full-IP deployment, says Ken Cavanaugh, director of business development for General Bandwidth, whose G6 can serve as a reverse gateway for operators that opt for PacketCable's Line Control Signaling (LCS) infrastructure.

"Operators don't need all of that [VoIP] expertise right away," he says.


Whatever the level of optimism, most see 2003 as a year that will be populated with limited deployments, followed by a bigger and better push toward deployments in 2004 and 2005.

"I think you'll see some pretty good, scaled field trials at the end of the summer and the beginning of the fall this year with multiple MSOs," says Joe Matibag, director of product marketing with Nuera Communications. "But you'll see some POs (purchase orders) cut in the first quarter of 2004."

But sooner is certainly better, especially for vendors that are betting big on VoIP, but growing frustrated with slow MSO adoption.

"If cable operators don't come to the table and play, the vendors are going to quit," warns Stan Brovant, vice president of marketing for Arris Broadband. "If cable operators don't project that they'll be launching services and buying this equipment, then the vendor community can't justify further investments."

Although MSOs such as Comcast, Cablevision, Charter and Cox are doing a lot in this area, it's still not enough, as some vendors see it. "I need more operators than I see right now," Brovant says.


More gear makes the PacketCable grade

After approving the first MTAs and CMTSs for PacketCable 1.0 late last year, CableLabs retained that momentum at the end of certification wave 25, awarding qualification to call management servers from Cisco Systems (the BTS 10200) and Syndeo (the Syion 426).

The CMS, a central piece of the distributed PacketCable network, manages and maintains the call state for VoIP services through its two primary components: a call agent and gate controller.

Despite some recent achievements, there still remain several PacketCable components to be tested and approved.

How soon those other pieces are added to the testing mix is largely up to the vendors, but CableLabs is expected to begin testing media gateways and media gateway controllers in certification wave 26, which kicks off May 12.

Comcast prepares to rev-up 'Rev 0' architecture

Many sets of eyes are keeping close watch on Comcast's enterprising VoIP trial near its hometown of Philadelphia.

For the initial phase of the trial in Coatesville, a town about 40 miles Northwest of Philadelphia, Comcast is serving about 75 "friendlies" via an overlay network populated by a separate RF channel, CMTS and customer premise devices.

The next stage of the trial will involve an architecture that Comcast calls "Rev 0" (see chart above), a multi-headend, distributed PacketCable-like network that leverages DQoS. That architecture, combined with DQoS, will allow Comcast to run voice and data applications via the same network, but ensure that voice packets carry the highest priority.

"In our view, the heart of the compelling economics of VoIP is the integration of the plant, with the same RF channel and the same device in the home supporting both high-speed data and voice service," says Sam Chernak, Comcast's vice president of VoIP.

In addition to adding DQoS sometime this summer, the move to Rev 0 will also expand Comcast's VoIP footprint to about 180,000 homes passed when it "flash-cuts" to the area's fiber ring. When that happens, Comcast will offer the trial service to hundreds of employees in the area.

They won't pay for the service, but Comcast will oversee a gauntlet of back office testing, including billing cycles, provisioning and customer care. Those elements will have to be in perfect working order before Comcast even considers pulling the trigger on VoIP as a real business.

"The plan is to emerge later on in the summer with an overall product and service that holds together so that if in fact we wanted to do any kind of launch, I can sit up and say: 'Hey, this thing will work'." Chernak says.

Comcast has grand visions for VoIP, "but it's a question of getting it right and scalable," he adds. "It's very important for us not to stumble out of the blocks with something as critical as voice service."

Liberty Cablevision takes the host approach

Comcast is taking the lead on virtually all aspects of its VoIP trial in Pennsylvania, but Liberty Cablevision in Puerto Rico is taking a much different path, opting instead for the hosted approach.

The "host" in this case is Net2Phone, which, like Liberty Cablevision, has corporate ties to Liberty Media.

Currently in phase II, the trial is delivering service to about 200 paying subscribers. Liberty Cablevision is offering a per-minute "Every Day" plan and a flat, monthly rate under the "Celebrity" plan.

About 70 percent have opted for the Every Day plan thus far, says Mike Pastor, vice president of Net2Phone's cable technology division. Regardless of the tier, Liberty Cablevision's VoIP customers are using the service an average of 800 to 900 minutes per month.

According to recent measurements, Liberty Cablevision's call quality has also rivaled that of state-owned incumbent Puerto Rico Telephone Co. (PRTC).

Citing recent Mean Opinion Score (MOS) measurements, which average the voice quality of phone calls on a scale of 1-to-5, Pastor says Liberty Cablevision has averaged a 4.2–well above the 3.8 score required for a toll-quality call.

Liberty Cablevision was expected to complete Phase IIa by the end of last month. The MSO will then take the cap off to service a potential base of customers numbering in the thousands. Then the pilot will evolve into "a real production, VoIP residential phone service," Pastor says.

Marrying voice and video...without network QoS

Although the videophone has made a name for itself in boardrooms and as a tool for journalists covering events in the Middle East, a more affordable videophone for consumers may be closer than some might think. In fact, it's here now and ready for the cable industry, according to Motion Media Technology.

By hooking into existing cable modems, Motion Media says its mm146 videophone can render high-quality video (up to 30 frames per second) and audio without the need for network-managed QoS or an underlying PacketCable infrastructure. Instead of requiring the QoS offered by DOCSIS 1.1 or 2.0, Motion Media answers that question via a proprietary protocol it calls "Packet Buddy."

Rather than leveraging a "heavy" corporate-class video conferencing standard such as H.323, which is difficult to configure, Packet Buddy borrows only elements of H.323 and supports a string of phone compression standards (G.723, G.728, G.722, G.711 and G.729). PacketBuddy "is designed to be lean and mean, and not waste valuable bits," says Motion Media CTO Stuart Ross.

The mm146 also is designed to adjust for changing network conditions, and to compensate for packet jitter and packet loss automatically.

Even if network conditions turn extremely nasty, the device always maintains "lip-sync" because audio and video information are always sent within the same stream, Ross says.

"Can it compensate for non-service? No. But it can dance through the rain drops without getting wet," he adds.

If the device was sold through retail, the appliance would go for about $499 MSRP. But cable operators would likely pay much less if they offered it to their customers as part of a bundled service or subscription plan.

"They think it could be a killer app for them," Motion Media CEO Garey De Angelis says of recent discussions with cable operators. "It's something that satellite can't provide to their customers."