The previous generations of Docsis (3.0, 2.0, 1.1 and 1.0) could share bandwidth (though cable operators did need to free up capacity for Docsis 3.0 channel bonding), but that won't be the case for Docsis 3.1, which will instead rely on tiny orthogonal frequency division multiplexing (OFDM) 25KHz or 50KHz-wide subcarriers placed inside a larger block of bandwidth.

The good news is that bandwidth used for Docsis 3.1 will be about 50 percent more efficient than what's used today thanks to the elimination of guard bands and the introduction of low density parity-check (LDPC), a Forward Error Correction (FEC) scheme that eats up less bandwidth than the current Reed-Solomon approach.

A new class of hybrid modems will be able to support both Docsis 3.0 and 3.1 traffic, so that's another bit of good news. But the new channelization scheme for 3.1 means cable will need to free up space yet again.

At some point, cable could end up trying to fill up 200MHz-wide blocks with those aforementioned subcarriers for Docsis 3.1, but cable's engineers have decided that the table stakes for Docsis 3.1 will be 24MHz of spectrum. While that doesn't sound like a lot, operators will still need to free it up, as it's unlikely that they happen to have that much capacity just sitting around unused.

But why is 24MHz the magic number? I ran that question by Cox Communications Inc. Senior Director of Network Architecture Jeff Finkelstein at Light Reading's Cable Next-Gen IP Strategies event in Denver last week. It turns out that part of it has to do with simple math: 24MHz offers a "logical common denominator" for the carrier widths used now for North American Docsis (6MHz-wide channels) and EuroDocsis (8MHz-wide channels).

Check out the LRTV interview for more on that, as well as Jeff's thoughts on the key benefits of Docsis 3.1 (beyond the promise of more speed) and potential migration strategies.

— Jeff Baumgartner, Site Editor, Light Reading Cable