I attended the recently concluded SCTE Cable-Tec Expo and came away impressed with the amount of system and component development activity in this industry. What started as a solution for homes that could not get adequate over-the-air broadcast TV signals, has become intertwined with cable-based broadband data services as service providers look to capture an increasing share of the “triple-play” of voice, video and data. The lines of distinction among the service providers is also getting blurry as traditional voice operators like AT&T and Verizon in the US offer broadband and TV services and cable operators like Comcast and Time Warner Cable offer broadband and VoIP voice services, in addition to the traditional TV offerings
The convergence of triple-play services also brings the challenge of designing networks that are scalable enough to deal with the explosion of data consumption. Several of the presentations referenced “the Nielsen Curve” that plots the maximum downstream data rate over time. This curve shows a very steady 50% yearly increase from the 300 bps performance in 1982 to 12Mbps in 2007. A couple authors were arguing that we are now in an era of “wideband” cable modems and the data rates have actually been increasing faster than 50% for the past few years!
There are several ideas about the proper architecture (or mix of architectures) required to handle this 50% (or higher) increase in downstream data rates that is being fueled by video, higher internet data speeds and more TV channels. The other aspect of the architecture challenge is what to do about the return or upstream signal. As user experience becomes more interactive, the data requirements increase on the upstream path, as well.
The new architecture ideas include running fiber as deep into the network as possible. While this solves the bandwidth problem, several of the large MSOs pushing this approach are having a hard time making the business model work, despite healthy numbers of subscriber additions. Networks that use EPoC (Ethernet PON over Coax) and DPoE (DOCSIS Provisioning of EPON) offer transport advantage over more traditional HFC networks without the cost of “fiber deep”. The industry is discussing changing the traditional 5-42 MHz return path bandwidth and in what portion of the spectrum to locate this additional bandwidth. In conjunction with this discussion invariably comes the discussion to extend the band to 1.2 GHz, from thecurrent1GHz (or below) physical plant. Finally RFoG (RF over Glass), where RF signals are transported over fiber (allowing the MSOs to maintain much of their HFC headend and back office equipment is still being touted as a viable option. Also in the mix are network concepts like SDV (switched digital video), node-splitting and hybrid gateways.
So, there are many options available to get to a stated goal of 2.5Gbps downstream and 1Gbps upstream cable data rates. There were a couple of presentations that did modeling to show that with the proper mix of existing technologies (node-splitting, SDV, hybrid gateways and the appropriate phase-out of analog channels), operators can accommodate the expected high-speed data, video-on-demand, HD and IP video) for the next decade with a 1 GHz physical plant! The dilemma for operators and equipment manufacturers is to determine the appropriate mix to satisfy their particular mix of SD, HD and analog TV channels, as well as accommodating broadband data rates and planning for DOCSIS3.1 and the eventual move to an all-IP network: a difficult choice indeed!
However, other than making for a dynamic environment in an industry known mostly for stability, why do we care? How the architecture discussion plays out will have major implications for compound semiconductor opportunities in this market segment. As fiber goes deeper into the network, whether this is directly through fiber to the premises deployments or increasing use of PON architect, fewer RF components may be required. Advances in GaN technology by companies like ANADIGICS, RFMD and TriQuint may also limit the compound semiconductor opportunity in this market. In addition, how does the industry answer the frequency split/extension question? Many of the return path products are designed to meet the 5-200 MHz option being discussed and the downstream path components routinely have bandwidth above 1 GHz different solution will create more development work.
Stay tuned for more in-depth analysis of this market and I do look forward to the days of 2.5Gbps downstream data rates!