Comcast's Jorge SalingerIs anyone ready for expanding the spectrum on their HFC networks to 1.7 GHz? DOCSIS transport at 1 Gbps? QAM modulation at 4,096 channels?

Not by a long shot. But experiments have demonstrated that all of it is technologically feasible, which would make HFC extensible for years. The inevitable, ultimate transition to fiber-to-the-home won’t be necessary for a couple of decades.

Even as the cable industry was learning to move at Internet speed and respond to the market in months rather than years, it’s been giving plenty of thought to how to gradually evolve the HFC network over the next decade.

Two sessions on Tuesday addressed the issue directly: “HFC 2.0: The evolution of architecture” and “The gigabit network: New possibilities in HFC.”

Kevin Leddy, executive vice president of technology policy and product management at Time Warner Cable, set the table at the latter. Operators can always do node splits, he said, but each generation of splits is at least twice as expensive as the previous one – that solution doesn’t economically scale.

You can reclaim analog bandwidth and remove analog elements from the network. You can bond more and more DOCSIS channels, but that might drive the prices for CPE through the roof.

That’s on the downstream. On the upstream, operators are eventually going to run out of capacity.

So what are cable operators to do, Leddy asked, knowing that he had a panel full of people who’d been thinking long and hard about just that question?

Jorge Salinger, vice president of access architecture at Comcast, said there are several things to consider to open up more bandwidth on the downstream, including moving QAM generation downstream. In other words, move QAMs out of headends and put them in hubs or nodes. That keeps transmission in the IP realm longer.

As a companion measure, there are advantages to moving the PHY, and perhaps PHY and MAC both, deeper, as well.

“This can be done for the CMTS, for edge QAMs and for CCAP, he said, suggesting that doing so makes the most sense for CCAP. The upside is simplifying transport at the headend and moving to baseband lasers.

On the other hand, that path would lead to putting many, many more sophisticated pieces of equipment out in the field, “which concerns some people,” Salinger said.

Improving MAC and PHY are possible. Comcast has tested 1024-QAM, and it works, Salinger said. Going to 4096-QAM thus becomes that much more likely.

Another improvement would be the use of more efficient forward error correction (FEC) codes, with a switch from the Reed-Solomon (RS) codes widely in use today to low-density parity check (LDPC) codes.

He also suggested operating in currently unused spectrum above 1 GHz. “We might be able to get another 500 MHz, though this will require new equipment.” (Others speaking later suggested that cable might be able to go to 1.7 GHz.)

Next steps include enhancing DOCSIS and developing Ethernet PON over Coax (EPoC), “which to us is a great alternative to Metro E,” Salinger concluded.

John Chapman, CTO of Cisco’s Cable Access Business Unit, took the lead delivering some of the results of a massive, 185-page paper co-authored with Robert Howald, a member of the CTO office at Motorola Mobility, and Michael Emmendorfer, Arris’ senior director of solution architecture and strategy.

Chapman talked about improving the upstream, including expanding the current upstream allocation from 5 to 42 MHz up to 85 MHz (the mid split), expanding beyond 85 MHz to 200 MHz, or even 400 MHz (the high split), or some combination of those two with a separate allocation above 1 GHz (the top split).

“Logic suggests the mid split or the high split,” Chapman said. “We don’t like the top split – there’s just too much attenuation.”

For the short-term roadmap, he and his fellow authors recommend that cable operators make more efficient use of what they have available now, but eventually they should plan on expanding not just to 1.5 GHz, as Salinger suggested, but even beyond to 1.7 GHz.

Doing all of that, Chapman said, “we think we can get up to 10 Gbps on DOCSIS someday.”

Howald, expanding on that, said the upstream might get up to 1 Gbps.

“We know it will not be easy, but the capacity is there,” he said.

As cable operators move forward, almost every option they have will have some sort of effect on their noise budgets, Howald explained, which under some combinations of options might get taxed very badly.

One of the noisiest environments is, of course, the home, and the CMTS vendors suggested considering taking the home right out of the equation by introducing what they referred to as point-of-entry gateways – gateways that mount outside the home and insulate the HFC network from the noise in the home.

Ayham Al-Banna, systems architect in the office of the CSO at Arris, boiled the recommendations from the paper into a simple roadmap that began with a Step 0, in recognition that that particular recommendation is something that operators are already doing: use available spectrum efficiently.

Step 1 is additional node segmentation and splits, which he acknowledged gets too expensive.

Step 2 included a stew of options that include adding bandwidth, going for the high split, enhancing DOCSIS, adding new FEC schemes and moving to a new PHY – possibly OFDM.

Step 3, he said, is FTTH.

While the panelists were all in accord on many – though not all – suggestions for the roadmap, session moderator Leddy saw fit to remind the speakers about the literal bottom line on all they had to suggest.

“I’d like to see the price tag for our company,” Leddy observed. “It would be over $5 billion for us.”