In my last two posts of the Technical Guide to DOCSIS 3.1, we explored both the bandwidth capacity increases, and the technical advantages enabled by DOCSIS 3.1. Over the next few years, we will start to see wide-scale commercial adoption of DOCSIS 3.1. But what happens next? DOCSIS 3.1 still has some faults that will need to be explored to further advance the communications industry.
One of the challenges with Orthogonal Frequency Division Multiplexing (OFDM), the new multi-carrier transmission enabled by 3.1, is that clipping may occur due to high peaks in the waveform when applied with nonlinear amplification. Digital Video Broadcasting — Terrestrial (DVB-T) employs two techniques called Active Constellation Extension (ACE) in the QAM carriers and Tone Reservation (TR) as mechanisms to reduce the peak-to-average power ratio (PAPR). TR directly cancels out peaks in the time domain using simulated impulse kernels made of the reserved sub-carriers in the OFDM signal. ACE helps reduce the PAPR by extending outer constellation points in the frequency domain. As with forward error correction, there is a cost to use these techniques. In this case, there is a small average power increase and at most 1% served sub-carriers. These approaches are complementary, as TR is used in low order modulation while ACE is used in the high order modulations available in DOCSIS 3.1. In conjunction, they provide a gain of approximately 2dB.
The second major constraints of OFDM is that the channel bandwidth has to be more than the modulation rate in samples per second in order to prevent inter symbol interference. This is the orthogonal condition, which means that spectral efficiency improvements can only be gained by using higher order modulation at the cost of lower noise and nonlinearity tolerance.
Improving the Standard
Spectral efficiencies could be improved using a new approach called SEFDM (spectrally efficient frequency division multiplexing), which effectively compresses the subcarrier spacing while maintaining the modulation rate and system performance.
SEFDM can be viewed as an improvement over OFDM as it is adopts the same subcarrier approach but intelligently packs the symbols in optimal dimensional space, interleaving the OFDM signal effectively shrinking the spacing between adjacent sub-carriers. There are several papers which have shown between 20% and 25% improvement in spectral efficiencies with negligible impact on required SNR. All this does come at a cost, with increased complexity at the receiver and transmitter; however, according to Moore’s law, about every two years the computation power per unit price doubles. In other words, the computing power should be available to support this digital multi-carrier modulation method, and more. With that available processing power we will undoubtedly see improvements in the overall spectral efficiencies.
Additional improvements in spectral efficiency can be achieved by transitioning from traditional analog optics to digital optics in the fiber segment of the HFC network. Analog optics, utilizing amplitude modulation of the optical signal, has been used in the HFC network since the first DOCSIS deployments. By moving some, or all, functionality of the DOCSIS physical layer (PHY) to the fiber node as with R-PHY or D-CCAP architectures, the optical signal can utilize digital modulation, thus eliminating:
Digital optics have been shown to support higher order modulation levels due to the superior noise performance, allowing cable operators to make better use of DOCSIS 3.1 technologies. Looking even further into the future, certain CMTS and CCAP vendors have demonstrated that DOCSIS 3.1 technology can utilize an even greater amount of the RF spectrum available on the HFC network, well beyond what the current standards support. One approach to achieve higher throughput is to use R-PHY architecture and push the fiber node deep enough into the HFC network to eliminate all amplifiers in a N+0 configuration where the RF Coaxial segment becomes a last-mile drop to the home.
Brownfield vs Greenfield
Increasing bandwidth demands, gigabit pressures, aging infrastructure, increasing OPEX and decreasing ARPU is causing movement in fiber technologies for cable operators.
Optical fiber is the most obvious choice in greenfield deployments. There are several technologies that combine DOCSIS with fiber aimed at operators who have both pure HFC and pure fiber networks. One approach used by many operators is RF over Glass (RFoG), which allows operators to leverage existing investments in provisioning OSS systems, headend, and subscriber equipment. As such this is an extremely attractive approach while transitioning the plant to fiber. However, unless combined with DOCSIS 3.1 with higher modulation capabilities and larger spectrum utilization, or if leveraged for future optical node splits, the customer would essentially see the same bandwidth. RFoG in itself does not provide any additional enhancements besides lower OPEX, reliability, QoS and potential growth capacity.
To provide additional capacity, operators also have the option of using DPoG and DPoE — DOCSIS Provisioning over GPON/EPON. Both afford operators similar training cost savings while leveraging their existing DOCSIS provisioning OSS stack when deploying PON access technology, though they require new equipment at both headend and subscriber. DPoG will provide service at around 2.5Gbps downstream and 1.25Gbps upstream, while DPoE (using 10G-EPON) can support 10Gbps downstream and 1Gbps upstream. Similar to the DOCSIS standard, PON standards are rapidly improving, effectively doubling every nine months. Using time wavelength division multiplexing passive optical networking, TWDM PON (also referred to NG-PON2), operators can already reach 40Gbps using wavelengths that specifically avoid interference from GPON, 10G-PON and RFoG on the same fiber. Using a DOCSIS overlay, the ONTs appear the same as cable modems in terms of provisioning and setting up services.
Recent advancements in fiber have changed the deployment and architecture focus for cable operators. Rather than requiring a completely new physical infrastructure and logical architecture, operators can leverage the benefits of fiber using their existing infrastructure. In particular R-PHY and D-CCAP enable operators to feed a remote node with fiber to distribute the signal over existing HFC infrastructure to the home. When combined with increased spectrum utilization, spectral efficiencies and possible high order modulation of DOCSIS 3.1, digital R-PHY provides an interesting cost-effective means to dramatically increase network capacity while investing in a fiber backbone. PON is the endgame.
In closing, over the past few years the number of options for cable operators to achieve gigabit speeds has grown drastically. Some options are more applicable for greenfield while others are strategic investments for long term capacity growth in brownfield areas. CableLabs, ITU, and SCTE continue to innovate and benefit from each others’ developments. Because these innovations continue to make data communications more efficient and more reliable, the DOCSIS track has no foreseeable ending. Constantly improving the reliability and capacity at which bandwidth is sent to and from the customer premises is what continues to drive revenues up for communication service providers. These constant improvements and changing landscapes are a big reason why I value being a part of this adaptive and exciting industry.
Stay tuned for my next blog, where I’ll explore the forces that are driving more operators to adopt fiber-optic architectures, further increasing the speed at which bandwidth is sent to and from the customer premises.
For more details and insight into everything broadband, subscribe to our blog.