Will Anyone Switch On Cable's Upstream Booster?

Verizon Communications Inc. (NYSE: VZ)'s newest FiOS Internet tiers expose cable's relatively weak upstream, but Motorola Mobility LLC insists that operators already have access to technology that can help them pack on some upstream muscle. (See Cable's Upstream Gap .)

The technology is Synchronous Code Division Multiple Access (S-CDMA), an advanced physical layer for Docsis that makes channels in the noisy, nether regions of cable's spectrum (in the 5MHz to 15MHz range) usable for things like channel bonding. Motorola estimates that S-CDMA could help cable reclaim one or two 6MHz-wide upstream channels to create upstream speeds that can rival FiOS's, which can now top out at 65 Mbit/s. (See FiOS Speeds & Prices Take a Quantum Leap .)

S-CDMA isn't new; it's been part of the Docsis specs since 2.0 and the days of Terayon Communication Systems, which tried (and failed) to use S-CDMA as a differentiator when it was hawking cable modem terminations systems (CMTSs) and cable modems. Motorola bought Terayon in 2007, and is now cast in the role of S-CDMA champion. But, so far, operators have opted to use Advanced Time Division Multiple Access (A-TDMA), the other advanced physical layer defined by Docsis 2.0 that's not nearly as good at mitigating noise as S-CDMA. (See Moto Preaches Cable's Upstream Savior and Motorola to Buy Terayon for $140M.)

Although S-CDMA could be useful should the need arise, even Motorola acknowledges that no one has deployed it. Cox Communications Inc. is one of the few cable operators that has at least kicked the tires.

But the level of interest in S-CDMA has been rising during the last three months or so, insists Jeffrey Walker, director of CMTS marketing at Motorola, admitting that some of that renewed appeal is is attributed to a "competitive response" to Verizon's new tiers. He says S-CDMA can give cable operators at least one more upstream channel, and perhaps two, providing enough for upstream bursts of 100 Mbit/s to 150 Mbit/s. Motorola and Cox have demonstrated a Docsis 3.0 upstream of 400 Mbit/s via the bonding 12 channels to at least show what's possible under just the right conditions. (See Cox, Moto Test 400Mbit/s Docsis 3.0 Upstream.)

Motorola is also advocating that the upstream jump to 256QAM, which would enable a 6MHz-wide channel to pump out 40 Mbit/s, up from 30 Mbit/s using today's 64QAM technology.

Cable's already looking at how to dilate its thin upstream in a follow-on spec purportedly called Docsis 3.1, but it's likely going to be years before that project is complete with products that support it. The use of a "mid-split," already defined in Docsis 3.0, can widen cable's upstream to 85MHz, but that can't be implemented overnight, either. S-CDMA can be put into play now to help bring Docsis upstreams much closer to the performance being touted by FiOS, Walker says. (See The Docsis Addendum and Cisco Hints at What Comes After Docsis 3.0.)

And a Docsis 3.0 implementation of S-CDMA could aid both residential and business customers. As Walker explains it, a cable operator, for example, could bond four upstream channels for a commercial application and fuse together to others for residential services -- if the MSO was able to take advantage of all the spectrum in the range of 5MHz to 42MHz.

Walker won't say who else other than Cox has tested S-CDMA, but notes that several operators, including two "major" North American cable operators, are giving the technology a pilot run in anticipation of deployments.

Is a bigger upstream a big priority?
A cable engineer at a top five U.S. MSO admits that if an operator is looking for an incremental gain on the upstream, S-CDMA "is as good an option as any." But, here's the reality check, at least for this particular operator: while a speedier upstream would give the marketers something to toot their horns about, bulking up the upstream isn't a high priority at the moment.

Most of the bandwidth demand is still in the downstream, as evidenced by some new D3 silicon from Intel Corp. (Nasdaq: INTC) that will get North American cable close to producing bursts of 1 Gbit/s down when bonding 24 channels. The engineer says cable's upstream was growing at 30 percent a year in recent years, but it's slowed down to somewhere in the neighborhood of 18 percent. (See Intel's New Docsis 3.0 Chip Guns for 1-Gig .)

Upstream demand has apparently slowed down as the use of Slingboxes flattened out and P2P traffic began to disappear. "There's no big application out there to drive [upstream] growth. The only thing left is videoconferencing, and that's just not a significant issue at the moment," the engineer says, noting that operators can still do other things, such as split nodes, to create more upstream capacity.

While the new FiOS tiers apply some pressure for cable to do more in the upstream, it's still questionable if Verizon will attract many customers to its most expensive, speediest tiers.

Walker, meanwhile, is convinced that the demand is there for operators to adopt S-CDMA relatively soon. "They have to offer that upstream bandwidth," Walker says, predicting that full deployments will start in 2013. "They need it in the next 12 to 24 months."

But consider S-CDMA's relevancy to be on the clock. As another cable engineer put it, "the value of S-CDMA diminishes" once Docsis 3.1 comes into play. If 3.1 calls for a "high-split" that moves cable's upstream ceiling to 95MHz or even 200MHz, then "there would be adequate clean spectrum" and little need to fish for channels in the noisy bottom portion of the range.

— Jeff Baumgartner, Site Editor, Light Reading Cable

macemoneta 12/5/2012 | 5:28:27 PM
re: Will Anyone Switch On Cable's Upstream Booster?

"Most of the bandwidth demand is still in the downstream"

This is the definition of a self fullfilling prophesy.  It creates a catch-22; no one can develop applications for a high-capacity upstream, because cable companies don't offer it.  Cable companies don't offer it, because there are no applications.

I can imagine many uses for  symmetrical connection bandwidth, but that's as far as they get without a ubiquitous reality.

yarn 12/5/2012 | 5:28:27 PM
re: Will Anyone Switch On Cable's Upstream Booster?

If reliable then perhaps there's a use for business consumers but you probably end up cannibalizing on other higher premium services designed for that.

EuroCableGuy 12/5/2012 | 5:28:24 PM
re: Will Anyone Switch On Cable's Upstream Booster?

As a Euro-guy, this is maybe the right time to point out that EuroDOCSIS upstream channel band extends up to 65 MHz, providing a good 23 MHz (or 60%) more bandwidth than its US cousin...

A large part of the DOCSIS-NG white paper presented a couple of weeks ago in Boston is also dedicated to upstream improvement, so I guess it's at least perceived as a medium term priority by MSOs.

As much as a Catch-22 this is also a case of "build it and they will come": when broadband brought "always-on" Internet to every computer in the house, nobody imagined how different the use would be from the days when you had to dial-up to get online...

We can't really predict what a fat upstream will change in our lifes but it will definitely change the way we use our Internet access. A few ideas on top of my head:

* HD videoconferencing (with HD voice while we're at it)

* Cloud services, especially backup into the cloud: how long to backup your 1TB HDD with today's upstream? forget it...

* P2P OTT video distribution: yes Virginia, P2P is not only for sharing copyright infringing contents, it can help unicast video distribution to scale.

* [Something we can't figure out today, but will look obvious ten years from now]

fgoldstein 12/5/2012 | 5:28:23 PM
re: Will Anyone Switch On Cable's Upstream Booster?

The US should move towards Eudo-split or mid-split   The problem is the low split in today's plant.  It's there because the FCC has a stupid rule that a must-carry TV channel can demand to be carried in its broadcast slot.   Sicne broadcast begins at 54 MHz (channel 2), downsteram begins at 54.

But the DTV transition resulted in very few VHF-low channels (below 88 MHz), and with virtual channels, poeple don't even know what channel a station is on.  Cable boxes do translation too.  So the FCC should allow cable to migrate some low channels to upstream use, provided that any remaining VHF-low stations are accommodated via cable box virtual positioning.

The rule predates virtual channels, and I had assumed that it no longer operated. But a local OTA "shopper" channel on virtual 46 but OTA 10 (the only VHF in Boston) demanded placemetn on channel 10 on Comcast's cable boxes.  That's just inane. But it's no reason to cripple cable. 

Of course the transition will not be fast, as it requires every node and amp (if not at node+0) to be migrated to a switchable-split one.  CableLabs was told about th is over a deacde ago and chose not to act, but since they still don't have a clean FTTH path, not patching HFC is a problem too.

davidhoffman 12/5/2012 | 5:28:21 PM
re: Will Anyone Switch On Cable's Upstream Booster?

The real reason for Google's Gigabit project is to build a community wide affordable symmetrical very high speed broadband system to test out new web application with lots of real users.  Once that and other FTTH projects get established, the advantages of good upstream will be seen and demand should increase.

Sign In