A Billion Bits Per Second → Plenty of life in cable

Let's get real ... there is still plenty of growth remaining in a single 6 MHz channel. Using rounded numbers, the current limits are,

DOCSIS 1.0 = 30,000/ 5,000
DOCSIS 1.1 = 30,000/10,000
DOCSIS 2.0 = 30,000/30,000

As the effective throughput of 100Mbps Ethernet is actually only 90-something, all that DOCSIS 3.0 would have to do to achieve 90,000/90,000 would be to bond together three 6MHz channels in each direction. Even if there is additional unforeseen overhead in the bonding process, a reasonable worst case would only mean bonding four channels instead of three.

Using 256QAM on the downstream (which has been around since DOCSIS 1.0) yields a rounded value of 40,000. 30,000 represents 64QAM.

On a side note, I haven't been able to figure out the apparent infatuation with bonding 6 MHz channels. Why not increase the channel width? EuroDOCSIS currently runs 8 MHz channels. Cisco has quoted payload numbers of 51 Mbps for 256QAM and 8 MHz versus 38 Mbps for 256QAM and 6 MHz. Bonding channels, versus making the channels wider, sounds like something that someone would do to get around a patent.

Regardless of what DOCSIS 3.0 (or whatever) comes up with, it is pretty much a lock that both the downstream and return paths will be a shared service, akin to the current cable offerings. Here's an option that is not that much of a stretch from current DOCSIS and could beat fios at it's own game: give each customer their own 256QAM 8MHz downstream channel, their own 16QAM 3.2MHz return channel, and run all of that specific customer's services (data, on-demand video, and voice) on those channels.

Amen sir...amen

Not necessarilly to get around a patent. American NTSC TV channels have allways been 6 Mhz wide and systems are sesigned for this, whereas the Eurpean PAL and SECAM standards use wider bandwidth, since fewer channels are allocated there.

More likely, engineers are just used to thinking 6 Mhz. After all, it's amazing how sometimes the smartest people are unable to think outside the box that universities and societal programming have placed them in.

My cable assumptions:

Typical cable plant: 1000MHz

Typical # of homes running off the same cable plant: high 200's to 400

Even if each customer had their own 6MHz channel, 1000/6 means less than 200 homes could be served. As far as I know, splitting the cable plant into two costs a shit load of cash.

- Sherman

Oh come on...

Engineers are smarter than that. Think about all of the equipment that is currently transmitting on specific channels that are spaced at 6 MHz intervals. You can't move all of this gear around easily. It changes channel line ups, etc. All of the gear that is used to analyze the RF plant is also built with the 6 MHz channels in mind.

Aggregating the signals into larger than 6 MHz portions of spectrum is certainly interesting, but having the ability to place multiple portions of a data stream in different 6 MHz channels that are NOT contiguous is important for compatibility with existing systems.