Life After CCAP

Although Converged Cable Access Platform (CCAP) deployments are just barely underway, some of the industry's top engineers are already looking even further out on the network horizon.

In the short term, cable companies know they need to combine data and video management platforms and bring down the cost per channel for both data and video delivery. But, in the long term, they're starting to see the need for a much broader, more fundamental network transformation.

Indeed, the slow phasing out of analog cable and the growing demand for high-quality IP service delivery have cable operators exploring another major architectural shift. While over the years many cable operators have worked to centralize network operations, there is now growing interest in distributing intelligence deeper into the network and moving some of the functions that have resided in the cable headend down to the fiber node level.

"I think we're in a really interesting phase right now," says Jeff Heynen, principal analyst for broadband access and pay TV at Infonetics Research Inc. "A lot of these 10-year architectural discussions are happening right now for cable operators."

This decade-long view has operators considering distributed network architectures that could make cable infrastructure look more like the telecom variety. New network configurations are not likely to emerge quickly, but the distributed approach is being seriously evaluated even now. Engineers are modeling the possibilities, and cable operators are trying to determine what circumstances could create a compelling business case going forward.

Even as CCAP deployments start to gather steam, the industry knows that the converged access platform doesn't represent an end state for cable infrastructure. It's only one stepping stone.

Vendor viewpoint
Cable vendors Aurora Networks Inc. and Arris Group Inc. (Nasdaq: ARRS) are among the companies articulating a vision for distributed network architectures. Aurora was one of the earliest to promote a remote PHY offering; an approach that pushes the IP-to-RF conversion process down from the headend to the node site. The company calls its solution "Node QAM" or "Remote QAM."

Aurora's approach has at least one heavyweight supporter in Cox Communications Inc. CTO Kevin Hart. Although Hart hasn't made a commitment to any particular vendor products yet, he has stated his belief that a remote PHY solution, like the one advertised by Aurora, may make sense in select locations for Cox. (See Going Deeper With IP.)

Arris has been more circumspect in its approach than Aurora. But Arris's engineers have thought long and hard about the potential for different remote access architectures. Mike Emmendorfer, senior director of solution architecture and strategy for Arris, sees several options available.

First, Emmendorfer says, operators could move a small portion of the PHY layer, such as the analog-to-digital converter (ADC) or the digital-to-analog converter (DAC), down to the node but still leave a converter in the headend. Second, they could port the lower PHY to the node without leaving a converter in the headend, or move the full PHY layer to the node, which would shift signal modulation, forward error correction (FEC), and other functions downstream. Third and finally, they could shift the lower MAC or full MAC layer to the node, which would ultimately make the hardware at the node look something like a mini-CCAP box.

"Wherever there's a need to use digital optics, any one of these paths could be chosen," says Emmendorfer.

One of the issues with distributed architectures is that there's a trade-off between performance and complexity. When video processing functions get pushed deeper into the network, there's less of an opportunity for signal quality to erode before an end user actually receives the signal. However, operators also end up with more complexity at the node site. In some environments that trade-off may make sense. In others, the benefits won't outweigh the cost.

When and where
Beyond individual vendor efforts, CableLabs has also taken up the cause of studying what it calls Distributed Access Architectures (DAA). By looking to China, where cable providers have already experimented heavily with fiber-deep networks, the organization has begun to integrate new DAA solutions into official DOCSIS specifications.

CableLabs recently published the first version of a C-DOCSIS (the "C" stands for China) system spec, and it has formed an operator team to assess viable approaches for distributed network deployments. (A recent CableLabs presentation on the subject is available for download. See also CableLabs Goes Global.)

In the meantime, analyst Heynen believes that interest from US operators in distributed architectures is real, but he doesn't think we'll see anything for "a few years." He also believes the approach may appeal initially to those operators who see WiFi as a strategic service. "Those are going to be the ones that look at this distributed PHY architecture first," he says.

Where quality is paramount, advocates say, a distributed access architecture makes the most sense.

In the very near term, the topic of remote access networks appears on the docket at next week's SCTE Cable-Tec Expo show in Atlanta. One panel on Remote PHY includes top executives from Rogers Communications Inc. (Toronto: RCI), Comcast Corp. (Nasdaq: CMCSA, CMCSK), Cox, and Cisco Systems Inc. (Nasdaq: CSCO). These are some of the top minds in cable, and they're looking at a vision of cable infrastructure that goes far beyond the converged access platform.

So there is life after CCAP for cable. And distributed network architectures are very likely to play a big role in that life.

— Mari Silbey, special to Light Reading Cable

Light Reading tackled this subject in a recent ground-breaking webinar. In a session entitled "Virtualizing CCAP: Developing A Distributed Access Architecture," cable technologists from Cox, Juniper Networks, and Gainspeed explored the concepts behind a distributed access architecture and explained how that architecture would work. Click here to check out the webinar. We'll have a story on the webinar next week.

COMMENTS Add Comment
albreznick 10/18/2013 | 10:32:24 AM
Re: Distributed access architecture Hey, folks. The link to the Webinar is now at the end of the story as well. Definitely worth checking out.
albreznick 10/17/2013 | 3:31:00 PM
Re: Distributed access architecture here's that webinar link I wanted to share, Check it out.  https://webinar.lightreading.com/16542
victorblake 10/16/2013 | 1:01:12 PM
More like telco ? No. More like telco ? Not really. VZ architecture for example is highly centralized -- aka Super Head End. Same for ATT.

Article just doesn't make sense to me. First, it isn't clear what you mean when you say analog cable. The signals are still analog (AM or analog modulated laser) carrying DOCSIS. If by analog you mean analog video. Again, all MPEG encoded video content is still forward transmitted over AM as well. If by analog you mean "analog television" then the only cases of that have been (this is not new) for many years re-trans and public access carriage.

Second, cable operators are extremely highly distributed. Much more so than telcos. There are more cable hubs than CO's and LSOs for any given serving population -- by and far. Much much more distributed. In terms of services architecture, platforms for things like ad insertion and VOD are somewhat distributed. At the current time I would describe the average cable distribution of VOD as more distributed than the more centralized equivalents from ATT and VZ. That varies by geography, technology, and with other factors, but on average the VOD systems (xOD and start over) are serving smaller populations than their telco counterparts.

As to pushing the QAMs closer to the edge -- there are any number of compeeeting technologies that can increase throughput. Those include running Ethernet or EPON for example to remote QAMs (aka remote PHY) or increasing the throughput in the existing phy (ala 1024QAM). As long as the required specifications can be cost effectively met in the product, and the network is of sufficient physical quality, the higher throughput will win out over distribution of the gear because the cost of distributing gear is high (electrical power consumption in the osp, increased operational costs, etc.).

Back on telco vs. cable, you say that cable cos are going to decentralize more like telcos, but telcos are become more highly centralized. Eliminating RT's left and right and using FTTx (FTTN/FTTC) to reach deeper with fiber to eliminate remote equipment because of that same high opex. Likewise FTTP (FTTH/FTTB) also eliminates osp gear reducing opex. So again, what's happens appears to contradict what you are saying.
gconnery 10/16/2013 | 12:52:29 AM
More on Distributed Access "Top minds in cable".  That line just cracks me up!

Anyway, while I'm sure this is interesting to cable heads, a proper sequel to CCAP would be something like IP gateways in the home with say Microsoft-defined encryption that could be implemented by TVs and retail DVRs/STBs.  That would be a proper successor to CCAP.
albreznick 10/15/2013 | 5:41:37 PM
Distributed access architecture That's right, JT. Thanks for that reminder. I will post a link to that archived Webinar shortly so folks can check it out at their leisure. Definitely worth viewing. 
JonathanT231 10/15/2013 | 5:07:37 PM
More on Distributed Access Hey Mari - Alan B just moderated this webinar on distributed access architecture. Worth a look.
Sign In