Engineering-wise: Time for the Time Domain
It’s time for upstream engineers to start thinking in the time domain as well as the RF spectrum domain. We can use time domain data to maximize our upstream capacity and robustness as we load up the upstream with DOCSIS 3.0 carriers, turn on S-CDMA, and/or prepare for OFDM in DOCSIS 3.1.
The good news is that there are lots of options for how to get that data, all the way from using highly capable vector signal analyzers, through digital sampling oscilloscopes, including pocket versions that plug into laptops, and finally using a modern CMTS.
Here’s why we need time domain captures: as we load up the RF upstream spectrum with more DOCSIS 3.0 carriers, we push the limits of our laser dynamic range, especially older laser technology. This means that under fully loaded conditions, the sum of all those DOCSIS 3.0 signals and transient noise such as impulse and/or burst noise, stationary ingress, and even excessive collisions in contention mini-slots from all the increased data traffic, can cause the laser transmitter to clip and create non-linear distortion products across the entire upstream spectrum, including above the diplexor limit.
Up to now, the most common technique for detecting transient noise has been to compare the ‘real time’ upstream spectrum to a max-hold version from the spectrum analyzer; the latter shows how much transient noise is caught by the spectrum analyzer as it scans in frequency. Another technique is to examine the upper out of band spectrum for non-linear distortion products that come about when the laser goes non-linear. The problem with the first approach is that it only captures a portion of, or misses entirely, transient noise events. The second approach only captures events after they have driven the laser into nonlinearity. We need to detect impairments proactively before such problems occur.
For complete characterization of the upstream, a modern vector signal analyzer (VSA) is the best tool. Transient events can be triggered for and sampled in the time domain, and FFTs of the spectrum showing the entire event can be displayed, including out of band distortion products.
Further, an FFT of a time segment that is free of impulses will show the actual Gaussian noise floor against which an impulse-tolerant DOCSIS waveform such as S-CDMA or OFDM can operate; remarkably high orders of modulation are actually possible. We also need time samples to measure the impulse lengths in order to configure S-CDMA and OFDM optimally to trade off latency vs. robustness.
But VSAs are pricy, so we’re not likely to see them in every headend anytime soon. A digital sampling oscilloscope that covers the upstream is less than a tenth of the cost, and there are even pocket DSOs that plug into laptops via USB that do a decent job of capturing the upstream spectrum for under $200. I took one of those around the world several years ago and used the captures in a pocket arbitrary waveform generator to test DOCSIS A-TDMA and S-CDMA configurations below 15 MHz for example. Grab an upstream snapshot, play it out in the lab, test new equipment capabilities.
Plus, once you get the time-domain samples into your computer, you can use MathCAD or MatLab, or open source options like SciLab, Freemat, or Octave, or even Excel, to do the FFT analysis to see what is really going on in your upstream spectrum.
Go to our time doman web site to get any of these software packages, see a pocket DSO similar to the one I used, and read an example procedure for characterizing your true upstream noise floor.
But wouldn’t it be great to have this capability built into the CMTS itself? Luckily, the CMTS chip vendors already thought of it, and the most recent designs have separate processors just for doing captures and spectral analysis with no impact to the normal data processing in the CMTS. Someday, there will be a host of applications for modern CMTS and CCAP equipment for this purpose.
But while you’re waiting on a CMTS App store, the SCTE Standards program has formed a Special Working Group to help you characterize and prepare your networks for DOCSIS 3.1, optimize them for full DOCSIS 3.0 utilization, and develop new measurements to enable capacity and robustness improvements.
Topics already under consideration include upstream and downstream issues such as: how to more accurately characterize upstream transient impairments using a variety of methods and equipment as I’ve been describing; laser dynamic range optimization; identifying and minimizing nonlinearities in all-digital HFC networks; new DOCSIS 3.x carrier power profiles for the upstream; identifying and eliminating micro-reflections from poor termination, grounding, or cable damage; using leakage measurements that are below FCC regulations but that may still limit the capacity of the downstream; and a host of other conditions in our current networks that could be straightforwardly improved to pave the way for more DOCSIS 3.0 carriers, S-CDMA and OFDMbased DOCSIS 3.1 carriers. We’ll also identify any new test equipment that might be needed when DOCSIS 3.1 is deployed so that we can start training the workforce ahead of time.
A CMTS App store…now there’s an interesting concept.
Daniel Howard is senior vice president ofengineering and CTO of the Society of Cable Telecommunications Engineers