Why Cable Needs EPON Over Coax

Docsis 3.0 hasn't run out of gas yet, but the cable industry doesn't want to get caught off-guard when it does.

That's one of the key reasons several MSOs and cable vendors are putting some weight behind EPON Over Coax (EPoC), a proposed Institute of Electrical and Electronics Engineers Inc. (IEEE) standard that aims to bring PON speeds to hybrid fiber/coax (HFC) networks. (See EPON-Over-Coax Starts Its Standards Journey and Broadcom Crafting PON-Speed HFC.)

The current generation of Docsis modems can bond eight downstream channels and four upstream channels, enough for bursts of 320Mbit/s downstream and 120Mbit/s upstream. Down the road, those configurations could reach up to 32 channels, at least in the downstream -- enough for bursts of more than 1Gbit/s. Cable has demonstrated as much in some recent demos that stretched the technical limits of Docsis 3.0. (See Virgin Widens 100-Meg Footprint, Tests 1.5-Gig , Comcast Also Thinking Big With 1-Gig, UPC Tests 1.3Gig Wideband and The Ultimate Cable Modem .)

But there might be limitations to how much more the chips can support.

"We need to get ahead of this," says Jeff Finkelstein, senior director of network architecture at Cox Communications Inc. , one of EPoC's backers. "If we just sit back and wait for things to happen, it puts us at a disadvantage with competition, with chip manufacturers, and with the needs of the customers."

EPoC could give cable a fresh look at orthogonal frequency-division multiplexing (OFDM), a modulation scheme that will be considered for the IEEE standard. OFDM has already begun to find a home with wireless as well as with Multimedia over Coax Alliance (MoCA) , a speedy coax-based home networking platform.

Some see OFDM as a next-gen modulation scheme for cable, steering the industry away from 6MHz-wide channels (EuroDocsis uses 8MHz-wide channels) and toward smaller subcarriers (40KHz to 100KHz) that can push out more bits per hertz. Today's 256 QAM technology does eight bits per hertz, or about 42Mbit/s for each 6MHz-wide channel. It's believed that OFDM could let cable pursue higher levels of modulation, such as 1,024 QAM (10 bits per hertz) or 4,096 QAM (12 bits per hertz), if you want to look way ahead.

"Will we reach 4,096?" Finkelstein asks. "I don't know. There are people who thought we'd never reach 256 QAM, and some thought 64 QAM would be hard. The reality is that we did reach it."

Most agree that EPoC's initial focus will be business services, but it could end up in residential networks as well. One advantage EPoC supporters can tout is the ability to be deployed piecemeal, wherever it's needed. And one aim is to ensure that EPoC can plug into the Converged Cable Access Platform (CCAP), a super-dense platform that will combine QAM and cable modem termination system (CMTS) capacity. (See Comcast Gets Ready for CCAP and Cable Rethinks 'Modular' CCAP .)

EPoC might not be the only approach that the cable industry is eyeing. CableLabs , for example, is believed to be taking kind of the opposite approach to EPoC: an advanced MAC/PHY that would use very wide channels, rather than small sub-carriers. (See Cable Ponders Life After Docsis.)

— Jeff Baumgartner, Site Editor, Light Reading Cable

Page 1 / 4   >   >>
AESerm 12/5/2012 | 5:44:53 PM
re: Why Cable Needs EPON Over Coax

If the evidence of one visitor to a recent Chinese cable and telecom event bears out, then there is another very large market interested in EPON over Coax -- or Ethernet over Coax (EOC). That was the big buzz word at ICTC last October, according to this blog from Incognito's Stephane Bourque. 

Jeff Baumgartner 12/5/2012 | 5:44:53 PM
re: Why Cable Needs EPON Over Coax

Jeff Finkelstein was pretty adamant that the cable industry, in general, thinks that there's "a lot of life left in Docsis," so this is some long-term thinking at play here. The EPoC standard is probably still three years away, so no one is sounding the alarm bells or anything like that. Plus, it will be deployed on a  targeted basis.   But this whole what comes after Docsis discussion reminded me of a video interview I did with Rouzbeh Yassini, considered the "father" of the cable modem and the notion that cable needs to do away with its 6MHz channel model. JB



Jeff Baumgartner 12/5/2012 | 5:44:52 PM
re: Why Cable Needs EPON Over Coax

As OFDM goes, the thinking, as I understand it, as they'll be able to get more bits per herz using it and allow operators to use higher, more effective  modulation schemes, but if someone else here with some OFDM knowledge wants to chip in, I'd like to know a few more reasons why it's being considered here.

As for the architecture, the way it was described to me, there would be some points on the network and in the home where the bit conversions would have to take place. You'd still have an OLT but use a media converter so the signals could be run on the coax using OFDM (if that's what the standard ends up calling for); then at the customre premise you'd have another converter to put the signal back to Ethernet and do GigE inside the home.  Very simplified view, but there would have to be multiple handoffs and conversions  as the signals traveled along the HFC plant. JB


paolo.franzoi 12/5/2012 | 5:44:52 PM
re: Why Cable Needs EPON Over Coax


The bits per symbol are related to channel SNR and the ability of the receivers to distinguish noise from bits.  You can build narrower QAM channels if you wanted to (heck we built voice band modems with QAM in the 60s).  It really boils down to how much DSP you want to instantiate in what manner.  To get the higher data rates, you will need lots of little OFDM ones as I don't recall OFDM being much more spectrally efficient than QAM or QPSK or DPQSK.  Its all just physics + Shannon + Nyquist. 

I am just waiting for them to issue a 10Base5 spec....or even better a 10Base2.




Duh! 12/5/2012 | 5:44:52 PM
re: Why Cable Needs EPON Over Coax

Doesn't sound like the kind of major architecture change we were talking about a few weeks ago.  I'm still not clear why you'd base a future evolution on EPON MPCP rather than DOCSIS MAC, if it's really about PHY improvements and you're not going to rearchitect the HFC plant.

I'm also not understanding why OFDM for this application.  As I understand it, OFDM pays off in highly non-linear, noisy and multipath channels.  Which is not what I think of in HFC networks.  For coax, I'd think the DOCSIS folks' approach, looking for higher symbol rates with larger constellations with wider channels, might make more sense. 

What am I missing here?

RBR 12/5/2012 | 5:44:48 PM
re: Why Cable Needs EPON Over Coax

The Physical Layer has always been a false dichotomy that consumes inordinate amount of effort yet misses the true value for EPoC.  TDMA and OFDM can equally challenge for greater spectral efficiencies than the operator networks can sustain; just as DOCSIS 3.x could grow and bond as well as any seasoned data protocol.  This is not to say that the chip vendors are not very much wed to the selections made in this area.  Value to the industry from EPoC will derive from a common MAC and TC Layer with 802.3 and 802.1.  Advances in the Ethernet family of standards are enabling this renaissance of all things over IP.  Further, there is only one market for DOCSIS chip sets, only one industry supporting its development, yet the ecosystem supporting Ethernet is immeasurable.  Such was the way with SONET and all of those vendors.  Its blah and mundane but coalescing around existing 802.3 and the higher layers that call upon its controls is the imperative we missed out on 10 years ago whose return was inevitable.  DOCSIS can’t avoid the consequences of economics.

msilbey 12/5/2012 | 5:44:48 PM
re: Why Cable Needs EPON Over Coax

I'd love a look at a conceptual timeline here. Something that shows when CCAP kicks in (for some operators) and when EPOC comes into play after that.

Also, how about a sense of how deep operators should be pushing their fiber before playing with further channel modulation over coax? Is there any consensus on that?

And a pony. Can I get a pony?

paolo.franzoi 12/5/2012 | 5:44:47 PM
re: Why Cable Needs EPON Over Coax


An EPON MAC is NOT the same as an Ethernet MAC.  I am completely unclear on why folks think there is synergy there.



joanengebretson 12/5/2012 | 5:44:46 PM
re: Why Cable Needs EPON Over Coax

I've got to question how practical it will be to bond 32 channels. Won't operators have to give something up to free up so many channels? Can they do that without negatively impacting service?

paolo.franzoi 12/5/2012 | 5:44:46 PM
re: Why Cable Needs EPON Over Coax

Ummm...the higher layers talk to a GPON chip over a GMII they talk to an EPON chip over GMII.

Nobody uses the ATM style features of the GPON standard.  They existed pre-implementation but nobody uses them.  The basic difference outside the MAC is 3 wavelength support (if desired), some of the extended reach stuff, and the OMCI.  That is about it.  Other than that your ONT is presented with an Ethernet Interface.



Page 1 / 4   >   >>
Sign In