BlackbirdTom Cloonan of Arris is confident of remarkable speeds. (Video and transcript below). At CableLabs' 2015 summer event he predicted DOCSIS would reach 15 gig (shared) in a few years and 50-80 gig a decade later. 15 gigabits would require using 1.7 GHz and higher speeds would require going to 6GHz and more. Jeff Baumgartner at Multichannel calls this "supersonic DOCSIS." 

"Sound crazy?" Cloonan asks. "Maybe, maybe not." Higher frequencies attenuate quickly in longer cables. If you run fiber to the edge of the drop cable, the loss is modest. At CableLabs, they demonstrated a 6 GHz signal over 50 meters of drop cable. 50 gigabits, here we come.
DSL has already done this, going from 1.5 megabit HDSL in 1995 to near gigabit over 106 GHz in 2015, Cloonan graciously acknowledges.
Some cablecos undoubtedly will decide that if they run fiber that close they might as well go all the way. Coonan suggests others will make the same decision as AT&T, which is going with from the basement to 10M apartments. T is comfortable with 500 megabits each instead of the even higher speeds possible with fiber home.  
Baumgartner at Multichannel and colleague John Eggerton at B&C are reporters I follow closely because they break so many stories. ARRIS' white paper has more details.


Transcript lightly edited from YouTube captions.
Good morning. 10 gigabits per second, 15 gigabits per second those are the numbers right now that we're talking about. We plan to deploy DOCSIS 3 in this decade using 1.2 and 1.7 gigahertz. Exciting, It's going to happen and we're happy to be a part of it.

But if you're like me you may be wondering a little bit about what comes next. What will we do in the 2020 decade? What will we do in the twenty thirty decade. We might need a little more bandwidth and we've been thinking about that quite a bit. We think there are some good things we can do perhaps we can push DOCSIS up to fifty or eighty gigabits per second. Possibly 100 gigabits per second or two hundred gigabits.

Sound crazy? Maybe, maybe not. Let's take a look at a few things here. Here's a chart you've never seen before (laughter) Years along the bottom along the x-axis and the y axis contains the maximum bandwidth. It starts at 300 bits per second in the early days very humble and potentially going up 200 gigabits per second by 2030 or so. Now we overlay on this chart horizontal line that indicates that 10 gigabit per second limit that we talked about with DOCSIS for 1.2 gigahertz systems. You can see those two lines cross somewhere around mid twenties, That's good and means we've got a decent runway ahead of us.

It doesn't run forever. Eventually it runs out of gas, somewhere in 2020 time frame. The question is what will you do. What will we do as an industry? There are a couple of potential paths you can take. One of them is to abandon HFC  moved to newer technology. Seems like an Ethernet point-to-point or switched Ethernet are wonderful technologies. Many MSO's will use them. That's great and we are supporters. We're building products for that. We think [fiber] needs to be in our toolkit and should be in the toolkit.

But there's another approach we can take more controversial. That is to extend the capacity of the DOCSIS HFC network that we've been working with for over 50 years. There're some ways we've been thinking about doing that. If we could, for example, take that 10 gig limit shown in the slide here and move it to a hundred gig or 200 gig it pushes that green line up and you can see it crosses the line much further into the future. It could take us out to the 2030 and maybe 2040. It's kind of exciting as something that we want to look at with you today and see what is your take on this extension of the cable plant.

The honest truth is we got this idea little bit when we did a paper last year that looked at how the telcos have evolved. I know they're the evil empire. What they did is kind of amazing, I think. They took a hundred year old twisted pair cable copper cable. It was limited to about 1.5 megabits per second in 1995 and through magic they extended it to a gigabit per second broadband system running over that hundred year old twisted pair today. About a 700 times increase over a twenty-one-year period allows them to extend the life of their infrastructure allowed them to provide broadband services. Ok ma, a lot of improvement and a lot of flexibility on when they could choose to move to PON.

The question is can we do the same thing with HFC network? Instead of using twisted pair with its limitations try coax which has much fewer limitations. How do they do that though? They stretch the fiber. They made copper or copper portion that were much shorter in length. That reduced the attenuation . They also increase their spectra continuously throughout that 21 year period.

Then they did something that was a little bit counterintuitive. They actually decreased their modulation. They went down from 128 QAM to 64 QAM to lower. That's kind of interesting. Doing that seems a bad idea but actually working together those two points can do wonderful things for us. That's what we propose to do in the HFC plan.

So can we take that lesson apply it to our own systems? The answer is yes we can. We could, for example, take a system that has a CCAP chassis in the headend run over fiber system. But don't stop that fiber at the typical fiber node today. Run it all the way to the tap as on the top of this picture. Then might have four coaxes running off of it. Those coaxes might only be a hundred-fifty-feet. So not as much attenuation. Or you can also do it fiber to the curb where you have a lot of homes. 

You have a problem with upstream DOCSIS bandwidth but there are fixes for that now, You can also spread your wings and a distributed access architecture where you have a known base CCAP box. So you can use it in a lot of applications. But will it work?

Ayham Al-Banna proceeded to demo 6 GHz over a 150 foot spool of coax.