The Bandwidth Hunger Games

18 Jan 2013

Author: Dr. Robert Howald, Fellow of the Technical Staff

With the announcement of the DOCSIS 3.1 initiative at the SCTE Expo, the industry once again embarks on a technology refresh.  For MSOs, this comes on the heels of aggressive DOCSIS 3.0 deployment, preparing for CCAP introduction, and rapidly transitioning into a multi-screen video provider.  With so much going on, what’s driving this next step?  What are the possibilities?  How might they be achieved?

What’s driving the move is the simple fact that, as powerful as the HFC network is, it operates significantly below its theoretical capabilities.  This is a painful realization with the downstream under constant pressure to support seemingly endless IP data growth.  Of course, when engineers say “theoretical,” we’re predisposed to think “impractical.”  However, since DOCSIS’ inception, major strides have been made in closing the gap between the theory and practice of digital communications.  This came about through a few key technology concepts, as well as advances in the real-time computing power necessary to productize mathematical ideas that were once unimaginably complex to implement. 

How underutilized is HFC?  Let’s consider the downstream.  Some simplified math makes the case.  An abbreviated form of the well-known Shannon Capacity expression for high CNR, such as the downstream, is:

Capacity (Mbps) = Bandwidth(MHz)] * CNR(dB)] / 3.

So, for 6 MHz and 40 dB CNR, the result is 80 Mbps.  With 256-QAM, we achieve less than 40 Mbps, net.  Where is the rest of it? 

Well, first realize that cable’s downstream QAM standard is 15 years old, prehistoric in technology years, thus missing recent advances.  Second, it does not even define modulation beyond 256-QAM.  Even if we wanted to crank up the bit rate per 6 MHz, this capability is not available within the standard.  The QAM profile is fixed and does not go up or down without manual intervention.  Lastly, because it is a fixed profile with no ability to adapt, operators implement a large cushion of margin to minimize service outages and truck rolls.

What, then, are the possibilities? What about that 80 Mbps?  First, DOCSIS 3.1 puts aside 6 MHz channel thinking.  It is then better to ask how close we can get to the bandwidth efficiency implied – over 13 bps/Hz.  The answer: we ought to snuggle up quite close to it.  The “how” of this increased bandwidth efficiency boils down to three key components:

 

  • Robust, higher bandwidth efficiency QAM formats with adaptability to reduce the margin penalty and optimize channel use.
  • Modern forward error correction (FEC).
  • Architectural evolution such as fiber deep in the network and point-of-entry CPE.

With these attributes, it becomes realistic to target at least 4096-QAM.  Then, at 12 bps/Hz (ignoring overhead inefficiencies), we’ve introduced a profile 50% more efficient than the best we can do today, and are within the ballpark of fully exploiting HFC’s potential.  THIS is what DOCSIS 3.1 is all about – scaling up as the network allows that opportunity to do so moving forward.

Finally, the capacity expression also hints at how we might achieve the 10 Gbps objective outlined by MSOs at the SCTE Expo.  Note that there are two variables that determine capacity, each with significantly different implications.  For example, consider doubling capacity.  According to the equation above we can:

  • Double the CNR, in dB.  So instead of 40 dB, implement 80 dB.  Hmmm…there could be some obstacles there.  Such as physics.
  • Double the spectrum.  Not simple or inexpensive, but completely doable.  We are far from full use of the coaxial bandwidth.  However, there are significant ramifications to plant and legacy infrastructure. 
  • A combination of the two. 

If you have your #2 pencil in hand, the latter is the obvious “all of the above” circle to fill in.  We may not be able to double CNR, but we can certainly improve it with architecture evolution.  And, new FEC enables more efficient use of a given CNR.  However, the clear and direct linear relationship of bandwidth to capacity identifies the path to major capacity growth – new spectrum.  In fact, new spectrum is the only path to the 10 Gbps/1 Gbps objectives described by operators for DOCSIS 3.1.  Note that upstream expansion becomes necessary as well.

So, as should be the case, DOCSIS’ next steps depend on the objectives to meet, and these have now been outlined by MSOs.  While the future is always uncertain, the nearest-to-certain driver we can bet on is growing bandwidth consumption.  Nonetheless, history suggests that some bets still end up surprising us.  Such as the bet that DOCSIS 3.0 will be the technology that carries us long into the future!

Tagged with :

Posted in

All Posts