Progress is being made. At the CableLabs Winter Conference, Cox and Motorola announced that the companies sent 356 Mbps upstream in the 5 to 85 MHz path. The test, done at Cox's Las Vegas system, used Motorola's DOCSIS 3.0 RX48 return path receiver modulator and the BSR 64000 CMTS Edge Router. Floyd Wagoner, Motorola's director of marketing and communications for the Infrastructure Solutions business, says the test enabled more aggressive modulation on a wider range of channels.
The test combined many innovations. On the spectrum front, the companies opened the upstream all the way to 85 MHz. At the same time, capacity in the lower regions of the spectrum that to this point has not been usable was made viable by synchronous code division multiple access (S-CDMA) modulation, Wagoner says.
A Tricky Road
The Cox/Motorola test was a high profile example of an industry effort to deal with the return path. The road is a tricky one, and is greatly impacted -- and ultimately will be aided -- by the further growth of DOCSIS 3.0 and its channel bonding features.
Channel bonding in DOCSIS 3.0 is more complex in the upstream than downstream. In the downstream, channel bonding is a relatively straightforward exercise in statistical multiplexing, notes Jeff Finkelstein, Cox senior director of network architecture. The CMTS is entirely in control and transmits the necessary bits simultaneously in the available channels to customer premises devices that know they are coming.
In the upstream, however, the CMTS doesn't know who will want to transmit, when or how much data they will have. Therefore, each user must request and get permission to transmit at a certain point in time. In addition, the network is a mix of bonded (i.e., DOCSIS 3.0) and non-bonded channels.
"A lot more effort has to go in at the CMTS to make sure everyone gets a fair chance and (that) it's equitably distributed among people requesting," Finkelstein says. "The more people requesting, the more complex it becomes."
New Tools, New Knowledge
Scheduling isn't the only challenge. As the frequency range increases and the modulation schemes become more aggressive -- from 64-QAM all the way up to 256-QAM -- both shortcomings in the network and the possible limitations of the lasers come into play.
This more sophisticated technology will demand more of operators. "There are things you can't see with a lot of the tools today," JDSU Product Manager Jim Walsh explains. "You have to upgrade the tools, upgrade the technicians' knowledge base. It is not SNR alone. You have to understand MER, code word errors, BER," he says, referring to modulation error ratio and bit error rate. "It is a whole set of metrics you didn't have to know about or keep an eye on before. Now, when you are getting into linear impairments, it's a paradigm shift. The challenge is enormous."
Walsh says the most basic need is a clean network. Ingress and other poor design characteristics that slid by without causing problems in more relaxed days can now cause issues such as laser clipping. "I see it more than I had expected," he says. "I should have predicted it, but it came as a little surprise. It is almost inevitable that [when assessing a system] we see it in at least a couple of nodes."
The Motorola/Cox Las Vegas test featured a variety of implementations, Wagoner and Finkelstein report. The basic structure featured 12 channels -- six operating in the 5 to 42 MHz spectrum and six from 45 to 85 MHz. The higher path, Finkelstein says, used 256-QAM. Wagoner says new diplex filters were necessary.
"The most challenging part was getting the software to support the 85 Meg return," Finkelstein says. "The hardware was capable of it for the most part. Firmware from the chip manufacturer had to be integrated into the Motorola code, and that had to be integrated into the entire ecosystem."
The good news is that the industry is acutely aware of its traditional Achilles' heel, and its engineers think that they have the tools in place to handle the upstream challenge. Future considerations may include the transition from Fabry-Perot to distributed feedback lasers, says John Dahlquist, VP of marketing for Aurora Networks. Cox and Motorola used DFBs in Las Vegas.
That may be the next step as the industry methodically tackles a traditional problem that, if not confronted, could eventually impact its ability to compete. "I would say that there is a lot of untapped potential in cable itself," Finkelstein points out. "DOCSIS has a tremendous amount of capacity to take advantage of, and HFC has a lot of life left in it."
Carl Weinschenk is a reporter at BTR. Reach him at email@example.com.