DENVER -- Cable Next-Gen Broadband Strategies 2010 -- Comcast Corp. (Nasdaq: CMCSA, CMCSK)'s lab and field trials of upstream Docsis 3.0 channel bonding have produced "sustained rates" of 75 Mbit/s to specific service groups, Chris Bastian, the MSO's executive director of network architecture, revealed here Thursday on a panel focused on what's next for cable's wideband platform.

"The results look very good," he said of Comcast's attempt to bond up to four channels in the upstream.

The test results have been even better in more controlled lab settings, where Comcast has been able to create sustained bursts of 100 Mbit/s.

The speed variations in the lab and field activities are a result of the channel widths used to conduct the tests. In the field, Comcast's tests have been limited to the combining of two 6.4MHz-wide channels with two that are just 3.2MHz wide. In the lab, Comcast's been bonding together four 6.4MHz-wide channels.

Next to IPv6 adoption, upstream channel bonding is among the Docsis 3.0 features MSOs are putting on the front burner. The downstream channel bonding element of wideband is pretty much licked -- many MSOs, including Comcast, have launched 100-Mbit/s Docsis 3.0 downstream service tiers. UK MSO Virgin Media Inc. (Nasdaq: VMED) is already conducting a 200-Mbit/s market trial, with another set for later this year. (See Comcast Gets Bizzy With 100-Meg Tier , Virgin Preps 100-Mbit/s Broadband Launch, and Comcast Gets Its IPv6 On .)

Although recent trials indicate that upstream channel bonding is starting to work as advertised, Bastian noted that Comcast has not announced when it might offer that capability to customers.

The general belief is that some initial deployments may start to appear late this year, with a faster ramp-up coming in 2011. Japan Cablenet is one exception -- it launched a bonded upstream service last year using Arris Group Inc. (Nasdaq: ARRS) equipment. (See Japan Cablenet Swims Upstream .)

One challenge of upstream bonding is to find eligible channels to bond, since most capacity in the lower part of the spectrum (usually in the range of 5MHz to 20MHz) is unusable due to high noise levels.

One emerging remedy is S-CDMA (Synchronous Code Division Multiple Access), an advanced PHY that can blast through that noise. It can free up channels to beef up single-channel upstream speeds, or it can be applied to a bonded channel group. Motorola Inc. (NYSE: MOT) has been championing that technology lately, and has notched field trials with Cox Communications Inc. and still-unnamed MSOs in South America, Europe, and Asia.

The vast majority of deployed Docsis modems and cable modem termination systems (CMTSs) use another advanced PHY layer called Advanced Time Division Multiple Access (ATDMA), which is less noise-resistant than S-CDMA. (See Moto: S-CDMA Starting to Spread.)

Bastian said Comcast is testing S-CDMA in the lab but has not made any commitments on field trials or deployment. He told Light Reading Cable that Comcast likely won't go beyond the lab until all of the MSO's CMTS vendors -- Arris, Cisco Systems Inc. (Nasdaq: CSCO), and Motorola -- reach S-CDMA feature "parity."

He estimated that an all-ATDMA upstream spectrum can yield about 100 Mbit/s, while a mix of ATDMA and S-CDMA channels might yield up to 150 Mbit/s.

Bastian, who is tasked with managing Comcast's CMTS roadmap, also shed some light on what the MSO is targeting ahead of the development of the Converged Multiservice Access Platform (CMAP), a dense, do-it-all box that represents Comcast's vision for the next-gen CMTS and edge QAM. (See Comcast Proposes Its God Box .)

Bastian noted that current integrated CMTS chassis support roughly 96 upstreams and 96 downstreams. Based on vendor roadmaps, Bastian expects CMTSs to support up to 48 streams per blade, bumping chassis limits to more than 200 downstreams and 200 upstreams, or a heavier mixture toward one of them.

— Jeff Baumgartner, Site Editor, Light Reading Cable