Cable-Tec Expo 2013 Round-Up: What’s Up, DOCSIS?
DOCSIS 3.1 was formally announced at last year’s Cable-Tec Expo. But once you’ve mapped out the technology roadmap for the next decade or two, it’s time to stop living in the past. Members of the gang that helped create CCAP and DOCSIS 3.1 have already moved on to the next thing. Ready for digital forward links, remote PHY, and the next next-generation architecture network (N2GAN)?
No? Of course not. CCAP and DOCSIS 3.1 haven’t even begun getting deployed yet. Heck, the D3.1 specification was only just published on Halloween, but Comcast’s Jorge Salinger, Cox Communications’ Jeff Finkelstein, and Cisco Systems’ John Chapman have already anticipated where the bottleneck is going to start constricting 5 years from now, and at Cable-Tec Expo 2013 they proposed a solution.
The aim of DOCSIS 3.1 and CCAP are to increase both the speed and capacity of the IP portion of the HFC network. They will – largely in the headend. But with time, as more and more video gets transmitted on the DOCSIS channel (and more of what gets transmitted is in HD), the analog forward link in the hybrid fiber coax (HFC) network will not be able to keep up and will become a gating factor.
Replacing it with a digital forward link will enable more cable modems to run at the higher-order modulation profiles included in DOCSIS 3.1. The move suggests getting the RF elements (the physical layer, or PHY) out of the headend and putting them into nodes Chapman, CTO at Cisco’s cable business unit, observed, “the move to digital drives the decision to remote PHY, not the other way around.”
Remote PHY is not, in fact, a new notion.
Chapman claims it is rooted in work that he originally did in 2001, and was tied in with the modular CMTS (M-CMTS) approach considered by CableLabs in 2004 as it was working on iterations of DOCSIS.CableLabs added the M-CMTS approach as the Modular Headend Architecture (MHA).
There were several elements offered for DOCSIS at the time, but two are relevant when talking about remote PHY. One was DEPI, the Downstream External PHY Interface, and the other UEPI, the Upstream External PHY Interface. The former was adopted into the DOCSIS spec, but with no market foreseen for the latter, it was not. Chapman says there is now work going on to have UEPI finally published.
Remote PHY is defined as the combination of DEPI and UEPI, plus a new timing specification now being planned that will work over longer distances and will define the external QAM channels for MPEG video, according to Chapman. CableLabs is evaluating this combination, which is being referred to as MHA version 2 (MHAv2).
Ten years ago, there was no market application for UEPI. In five years or so, there should be.
During the Remote Phy session at Cable-Tec, Finkelstein, Cox’s executive director of strategic architecture, described the impending problem, starting with a graph that mapped broadband speeds against time.
He noted that the maximum speed tracks directly with the weighted average speed for each user. The maximum speed drives the bonding group size, while the weighted average speed drives total bandwidth need (see accompanying table).
“That lower line is essentially cost,” Finkelstein said. “You move it by 1 or 2 megabits per second and that equals a significant increase in dB – dB in this case meaning ‘dollar bills’.”
Both lines have been on a straight trajectory, and there is no reason to think their direction will change. According to the chart, some cable operators are going to need to start going to 1 Gbps max speeds within 5 years.
CCAP will get you to headends with 360 service groups, and access networks will evolve from node +6 configurations to node +0, with each step requiring the addition of more QAMs. The largest cable operators might end up with systems that have 100,000 nodes.
“You can manage going from +6 to +1, but the hockey stick is when you go from +1 to +0,” Finkelstein said. The number of QAMs required might not fit in many headends. At that point, you have a huge scaling problem.
DOCSIS 3.1 and CCAP will get the industry where it needs to go, extending the utility of the HFC network for a very long time, but eventually we’re going to have to go to the N2GAN, Finkelstein said. The term and its acronym are somewhat tonguein- cheek, but still.
Operators have options: 1) Maintain RF in the headend; 2) install digital to analog converters in nodes; 3) Remote lower PHY – splitting the PHY between headend and nodes; 4) Remote PHY; and 5) Remote the entire PHY and MAC.
Salinger, VP of access architecture at Comcast, in his paper dismissed the first and last options. The second through the fourth are essentially variations, each successive one putting more intelligence deeper into the network. Salinger said he expects the fourth option, a remote PHY, will require neither. And as an MSO splits nodes, going digital to the node should be no more expensive or complex. “Remote PHY seems to me the best option,” he said. “Time will tell.”
There are downside considerations to pulling everything that’s RF out of the headend and putting it in the field, Finkelstein noted, among them the difficulty of figuring out how to manage such a network, but the advantages include enabling higher speeds, the ability to use Ethernet to reach remote units, backward compatibility, and it’s easy to scale.
“The headend simply becomes an Ethernet aggregation point,” Finkelstein noted. “And instead of building bigger and bigger boxes, you build smaller and smaller boxes.”
That would be coupled with a digital downlink. A digital downlink will end up having substantially lower MER (modulation error rate), needed to support the move to higher-order modulation schemes, including 1K, 2K and 4K QAM, Salinger explained.
He also thinks that an MSO can gradually migrate to a digital forward link as it segments nodes, which “should not be more expensive or more complex.”
Chapman added to the list of good reasons to install digital forward links and remote PHY: cable operators in the rest of the world are moving that way already; fiber is a more efficient signal carrier; you can get more wavelengths and more capacity on fiber; the downward trend in digital costs has been consistent.
Of course, if putting more intelligence deeper into the network is a good thing, why not just push CMTS functions out as well? “CMTSs are not easy to manage,” Chapman said. There are also security issues. In headends, CMTSs are largely secure; in the field – in nodes – they’d likely be more exposed to hacking.
“The proposition of remote PHY is that you can take a chip out of the CMTS and put it in a node. It means centralized software.
It means we’re not reinventing the CMTS,” Chapman said.
And here's something you might not have heard before. If everything to the node is Ethernet, that might be the end of DOCSIS.
And if remote PHY doesn’t lead to killing DOCSIS outright, it might tightly confine DOCSIS to the part of the path between nodes and subscribers’ homes. “Either DOCSIS becomes a dinosaur,” Chapman said, “or we’re going to have to start a Save DOCSIS Foundation.”
Take your time setting up a collection box though. You have 15 years, maybe 20, maybe more. The industry still needs to deploy DOCSIS 3.1, the subject of a pre-conference seminar.
At the seminar, CableLabs’ Matt Schmitt, director, DOCSIS, announced that the specification would be published before the month was out (the organization published the spec on October 31, making its promised deadline with hours to spare).
“We made a fairly bold assertion in October of last year that we would have them substantially complete and publicly issued by the end of 2013,” Schmitt said. “This is quite a bit faster than we have ever pulled off before.
It’s not a small project to do a new DOCSIS with a new physical layer underneath. It was an industry-wide effort and I tell you what, they’ve been busting their tails.”
CableLabs and its host of vendors also met the four objectives that were laid out at last year’s Expo. The first was making sure that DOCSIS 3.1 could scale to 10 Gbps on the downstream, and up to 1 Gbps on the upstream, while also supporting larger spectrum bands.
The second was lowering the cost per bit while also increasing network efficiency. Overall, DOCSIS 3.1 has 50 percent more data capacity over the same spectrum than its predecessor.
The third goal was making sure that 3.1 was backwards compatible with the previous versions of DOCSIS so that cable operators didn’t have to run two different systems or replace existing equipment.
Finally, the DOCSIS 3.1 specifications had to work on the existing HFC plant while supporting more capacity on the upstream and downstream. When there are market drivers, 3.1 can grow with the MSOs, Schmitt said.
“We believe we achieved each of our high-level objectives,” Schmitt said.
One of the reasons that the cable operator industry was able to pull off the fifth iteration of DOCSIS so quickly was the choice to use existing technologies from other fields.
DOCSIS 3.1 features a new modulation scheme, along with a more sophisticated forward error correction (FEC). DOCSIS 3.1 will rely on orthogonal frequency division multiplexing (OFDM), used in DSL, LTE, Wi-Fi and MoCA, among other technologies.
It will also replace the commonly used Reed-Solomon FEC with far more efficient low-density parity check (LDPC) codes.
Following Schmitt, Terry Cordova, Suddenlink’s CTO and senior vice president, moderated the “MSO plans and perspectives” panel. Now that the specs are almost in hand, Cordova asked the panel when cable operators planned on deploying 3.1.
Peripatetic pair Salinger and Finkelstein were on this panel as well. Finkelstein said the deployment schedule would be dictated by the silicon and system vendors.
“As soon as we can get to anything we can expedite it, whether it's CPE or headend equipment, we would gladly give up the lab time,” Finkelstein said.
Salinger said 3.1 trials would start next year with the hope of having actual 3.1 products available in 2015.
“This has been the fastest DOCSIS development ever,” Salinger said. “I think from the development of the spec, or development of the idea in 2008, to the availability of the product will be about half as long as 2.0.”
Time Warner Cable’s Howard Pfeffer, senior vice president, broadband group technology, said 3.1 capabilities would emerge on the customer premise side first, but would be pushed along by converged cable access platform (CCAP) developments.
[CED hosted a November 5 webcast that recaps some of the same material and provides additional details subsequent to the publishing of the D3.1 specification.
Presenters include Salinger, Finkelstein, Chapman, and Arris staff systems architect Ayham Al-Banna. The recorded 1-hour presentation is available for review (registration required).]
CCAP in action During Cable-Tec Expo, TWC announced it is beginning to deploy CCAP technology. The company is using both Casa Systems’ C100G cable modem termination system (CMTS) and the E6000 Converged Edge Router from Arris.
Casa took the wraps off its C100G CMTS earlier this year. Its integrated CMTS can support up to 8,000 downstreams when fully loaded. Arris announced the E6000 last fall and it became widely available earlier this year.
Comcast previously announced having conducted field deployments of CCAP while Cox Communications teamed with Cisco on modular CMTS trials earlier this year.
Cable operators have been pushing CableLabs, the SCTE, and the NCTA to work more tightly together for at least a couple of years. The introduction of the new DOCSIS spec was occasion for the two to announce they would work in tandem to accelerate the worldwide development and deployment of new cable network technologies.
That includes a framework that will enable parallel development of new technologies and supporting training resources by SCTE and CableLabs, including DOCSIS training and certification that begins this month.
The SCTE will begin to provide a DOCSIS Engineering Professional Certification (DEP) that certifies the knowledge of the interface requirements and operation for the deployment of DOCSIS devices over cable networks.
It will also oversee a training course designed to address such areas as Architecture, DOCSIS Layering, DOCSIS Network Operations and DOCSIS Technology Enablement in preparation for the DEP certification examination.
“CableLabs and SCTE each play important and complementary roles” said Tony Werner, executive vice president and CTO of Comcast. “By increasing integration of their core competencies, both organizations can enable the industry to capitalize on innovation faster and more efficiently.”
C’mon get appy
The Cable-Tec session “The evolving customer experience” provided an overview of TV apps, personalized recommendation engines, and cloud-based services. TV apps and recommendation engines are two ways that cable operators can keep subscribers from wandering away into the arms of another service provider.
AppCarousel managing director Terry Hughes said that the environment for apps in the video industry is still developing. The EBIF model of point and click with remotes never quite fulfilled its promise on the app front, Hughes said. Further complicating the TV app market are decisions on where a TV app should reside, which could include set-top boxes, Blu-way players, smart TVs, gaming consoles and tablets.
“Right how it’s a fragmented ecosystem,” Hughes said. “We have to fix that to get developers to the build the apps.”
Monetization could follow the same model as mobile apps with banner ads being placed on free apps. App users could leverage their existing relationships with Apple and Android stores to buy the apps.
“Apps are critical for monetizing the rich content that is out there,” said Hughes, whose company is working with Arris.
ThinkAnalytics founder and chief technology officer Peter Docherty said personalized recommendation engines increase subscriber stickiness and perceived value, increase VOD usage or buys, decrease overall churn and increase the customer base.
Successful recommendation engines incorporate user data from all the devices in a home. Docherty said having one way of making recommendations was akin to having just a hammer in a toolbox.
There also needs to be recommendations for each individual in a home in order to draw a bead on accurate viewing patterns.
Cox Communications is using ThinkAnalytic’s recommendation in its Trio guide. According to a survey by Cox, 76 percent of its users “strongly agree” that the recommendation engine made it easier to find and record movies and shows. The survey found that 56 percent of the users said the recommendation engine makes them aware of content they hadn’t previously known about while 32 percent said it made them watch more television.
Nagra’s Robin Wilson, vice president, business development, spoke about cloudbased services. “Just because something can be cloud-based doesn’t mean it should be,” Wilson said. “HTML5 is a good example of that. If someone tells you it’s a private cloud, I would be very suspicious. Private-cloud is a questionable definition.”
Using the cloud can lower capital expenditures and lower cost compute resources. The cloud can also be used to enable more features on legacy set-top boxes, according to Wilson.
Kit and kaboodle
The Reference Design Kit (RDK) was created to enable MSOs to introduce new video products and services faster than ever, by creating a means to cut the time it takes to design and manufacture set-top boxes and gateways from years to months. The RDK has been around for only a little more than a year.
The “All things RDK panel” at Expo delved into defining the RDK’s abilities, lessons learned from early RDK work and the support through the recently formed RDK Management.
Initially, the RDK was developed internally by Comcast by using best-of-breed opensource components from the Internet. It is now a community-based project that allows developers, vendors and cable operators to use a defined stack of software on one layer in order to provision set-top boxes and gateways.
The RDK allows all of the interested parties to develop once and then scale across multiple environments – in the CableCard/QAM/ MEPG-2 world of today, as well as in the IP environment of tomorrow.
In August, Comcast and Time Warner Cable formed RDK Management, LLC, which has taken over the licensing, training, community support and code management for the Reference Design Kit. In addition, the new entity will establish an expanded support program to provide technical support to RDK licensees as operators more broadly deploy the RDK platform.
Comcast has deployed RDK-based software in its X1 platform, which is the only deployment of RDK-based devices to date. At The Cable Show, Comcast announced its X2 software update, which will include the new RDK-based Xi3 box that is made by Pace and Humax.
Time Warner Cable is trialing IP hybrid boxes and a cloud-based guide in Syracuse, New York, with the goal of expanding the trials to Los Angeles and New York City this fall, but the RDK won’t come into play until the first half of next year.
Steve Reynolds, Comcast’s senior vice president of CPE and home networking said one area that has helped cut down the development cycle was using base ports.
“The good news is the process has gotten considerably easier,” Reynolds said.
“RDK 1.3 is the release version now. It’s relatively stable and a lot of the features are already working on it. With all of those SoC (system-on-a-chip) vendors already doing base ports on the silicon that is broadly used in the industry, it makes it a lot faster and a lot easier to get a port of the RDK running up on a specific set-top box.
“A base port is the first time that you bring the RDK up on a specific chip family. We call that the base port because it’s not necessarily fully integrated into the set top itself. We have those base ports on most of the common industry SoCs at this point and time. By using base ports it saves the OEMs months, if not orders in time, in bringing that stack on top of that set top hardware. The reason we started doing this whole thing in the first place was service velocity. By and large we’ve done that by getting these base ports out to the SoC community.”
Aside from Comcast and Time Warner Cable, other cable operators that have previously expressed an interest in the RDK include Charter Communications, Rogers Communications, J:Com and Liberty Global. Cox Communications is also kicking the tires. Cox’s Steve Calzone, director of video applications development, said his company was not yet a licensee of the RDK but rather an “evaluator.”
“We’re evaluating the technology to understand how it can really help us have faster time to market with applications,” he said. “What I know is it’s not the end-to-end cure all for our application development, but it’s a good piece of the infrastructure that we can use to really move a lot faster. Our intentions are to start developing with HTML5 to ultimately provide faster time to market for guide-based applications to our subscribers. We still need the other pieces to make it work to our subscribers’ benefit and the RDK is not that.”
Reynolds and other panel members stressed that the RDK was not an attempt to do a universal application platform, or a “Comcast jam down.” Steve Heeb, president and general manager, RDK Management, said those who have elected tp license the RDK could “vote with their code” on what it should be going forward. The point is that all of the members of RDK Management are able see the code as it develops, and decide for themselves whether it’s a good fit for them.
Heeb said there was a “huge opportunity” for vendors to be the next Red Hat or Linux by developing on the RDK framework, but “people in the community need to step up.”
Going forward, Reynolds said that instead of using Webkit, the QT element of the RDK was moving to Blink’s HTML5 implementation.
Also in the 2.x version of the RDK, OCAP will be optional instead of a core component.
“Our interest in the RDK is what we can do with next generation UIs moving to the cloud,” said Bill Warga, vice president technology, Liberty Global. “In our latest demo we are showing a re-skinning of the UI every five minutes. Every five minutes another page would load to show people that we were operating from the cloud and how quickly we can make changes. We’re excited about what the platform can bring us.”
The final panel of the opening general session at SCTE Cable-Tec Expo wasn’t so much about getting down to the brass tacks of cable technology, but more about giving attendees tips that they could take back to their respective jobs.
Part and parcel of creating a more nimble cable dynamic is providing employees, including field technicians, the proper tools to succeed, which in turn made them happier workers.
To kick the session off, moderator John Schanz, Comcast’s chief network officer and executive vice president, asked the panelists to talk about some of their companies’ biggest risks, or ways they embraced changes.
Cox’ Guy McCormick, vice president, engineering, recalled the risk his company took when it embarked on building its converged network, which he said was very cutting edge at the time.
Cablevision’s Yvette Kanouff, executive vice president, corporate engineering and technology, said one example of embracing change was Cablevision’s decision to join Netflix’s Open Connect program. Netflix’s Open Connect was launched early last year.
Netflix has said ISPs could receive its video directly at the interconnection point of the ISP’s choice, or closer to the edge of a cable operator’s network via its Open Connect Appliance chassis.
“It’s an example of ‘This is what our customers want so lets make it the best experience for them,’” she said.
Matthew Stanek, senior vice president of core, regional, metro and network operations, Time Warner Cable, cited the risks associated with the integration of the former Comcast and Adelphia systems in Los Angeles.
“More recently in Kentucky we took Insight and converted that into the Time Warner family,” he said. “It look a lot of effort and a lot of coordinatioin across multiple departments to move that along the lines of one network.
I think there’s a lot of risk to migrate to one structure and network.”
Marwan Fawaz spoke about his recent work at Motorola Home, which was formerly owned by Google prior to its purchase by Arris.
“You get to look under the hood a little bit and see what they were doing,” he said. “What was really fascinating to me was for every 10 projects they had that failed they would have one that would succeed.
These are projects that you don’t see.
I learned so much about learning from failures instead of focusing on success.
They grew up in that mentality.”
Fawaz said he also learned about “dogfooding” while Motorola Home was under the auspices of Google. Dogfooding allows each member of a company to test or trial a product or service prior to its release.
“From CEO on down everyone gets to try the product way before it's released,” he said.
“It starts in pre-alpha and the people know about the product. We need to do more dogfooding in our industry.”
Kanouff said the cable operator industry needed to embrace the fact that software is what runs the services instead of focusing so much on the network and hardware. Cable operators need to become more nimble with their services offerings, and embrace it as a mindset, according to Kanouff. .
McCormick said that while network convergence was great, network technicians and engineers needed to have tools and capabilities that were network centric.
Fawaz said he recently discussed GoPro cameras with Mike LaJoie, Time Warner Cable’s executive vice president and chief technology and network operations officer.
The small GoPro cameras could be clipped on a technician’s helmet. If he or she comes across a problem in a node the video could be viewed via the cloud by a supervisor to help fix a problem.
Schanz said a little over a year ago Comcast throttled back on the approvals needed to complete a job. Instead, employees that made changes wrote them in a log so there’s one large repository of all of the changes. Schanz said that dropped approvals by 58 percent and made employees feel more engaged and have fewer tasks to complete. ■