Pause a moment, if you will, to consider the plight of the oft-reviled cable television providers. At their roots, these companies operated essentially unidirectional networks of coaxial cable for distributing analog video signals to homes. That was then.
And this is now: those original co-ax networks have evolved into asymmetric, bidirectional digital data networks offering Internet access, voice telephony, security services—oh, yes, and high-definition television. They have become hybrid networks, with optical fiber between the head-end and the neighborhood fiber node, and legacy co-ax from there to your cable box, which is affectionately known as customer premises equipment (CPE). Every link in this chain is trembling under the strains of too many users demanding too much bandwidth in too many ways.
And consider the near future. Customers are “cutting the cord:” not disconnecting from the cable, but unsubscribing from the expensive TV services and instead taking their video through the Internet access. Wireless providers are muscling in, as younger customers eschew TV monitors, personal computers, and fixed-line phones in favor of their mobile toys. And 5G threatens to deliver broadband service wirelessly to every square meter of your property. All these trends threaten to divert revenue away from the cable providers. What’s a cable company to do?
Waves of Evolution
The industry’s response to these pressures varies from company to company. Some, constrained by lack of capital or by limited access to technology, are simply putting their head ends in the sand, as it were, and counting on all that installed co-ax to keep them in business. Others are rapidly transforming themselves into Internet access providers, application hosts, and even full-range fixed/wireless service providers. As Darwin might have suggested, mutation is occurring, and environmental pressures are sorting out the competitors.
These different transformations have disparate endpoints, but they all begin with the same sequence of steps: consecutive waves that will flood across the industry. Some providers will ride the incoming tide to its logical conclusion, while others will burrow in to escape change they can’t fund or can’t fathom.
Intel system architect Brian Kurtz summarizes the transformation in five steps: increasing bandwidth, distributing functions down into the network, achieving up/down rate symmetry, virtualizing network functions, and achieving a converged network. These sequential steps will take system operators through 2020, and should create platforms the operators can extend from there in the direction their business plans dictate.
The first step in the sequence, already in progress this year, is to boost the bandwidth of the existing hybrid fiber/cable network. Today, that network comprises a head-end, connected via analog optical fiber to neighborhood-level optical nodes (Figure 1 ). Each of these nodes interfaces a fiber to an electrical co-ax cable, which in turn daisy-chains through a series of local booster amplifiers, splitting off taps to homes as it goes. One co-ax run may serve as many as 500 homes.
The new data-over-cable service interface specification (DOCSIS) 3.1 leaves this network topology unchanged, but introduces wider frequency bands, higher-order modulation, orthogonal frequency-division multiplexing, and low-density parity check coding to boost the aggregate downstream bit rate on a fiber-and-co-ax run to 10 Gb/s. Up-stream data, in a nearby frequency band, bumps up to a maximum of 2 Gbps aggregate. These expanded bandwidths are shared by all those 500 houses that may unpeacefully coexist in a crowded service group.
DOCSIS 3.1 will require updating the media access controller (MAC) and physical- layer hardware (PHY) in the head end and in the customer’s cable modem or set-top box. But in general the network structure, broadcasting video and bi-directionally transporting Internet-Protocol packets between the head end and the customer, remains the same. Individual pieces of equipment or cable runs may have to change if they turn out to be damaged or incapable of supporting the new modulation schemes, but this is not on the whole a massive hardware upgrade.
The next step in this evolution requires a fundamental change to the network: from analog to digital fiber between the head end and the optical nodes in the neighborhoods. This is an enabling step—not so much to directly increase bandwidth as to improve the quality of service and to prepare the way for future steps.
Making the fiber connection digital opens many doors. It means that the providers can use inexpensive digital fiber and industry-standard Ethernet components. It means video signals can be digitized, packetized, and carried over the Ethernet protocol, rather than taking up big slices of dedicated bandwidth. And it means the D/A converters, PHYs for driving the co-ax, and often the Ethernet MACs will reside not in the head end, but out in the optical node boxes (Figure 2). Because of the PHY’s new location, these distributed networks are sometimes called remote-PHY or R-PHY configurations.
The R-PHY creates new possibilities, Kurtz notes. One really important one is full duplex (FDX), which enables symmetry of the link, unlike today where downstream capacity is 10X the upstream. At the same time the R-PHYs go in, cable companies will reconfigure from giant 500-house service groups and distributed amplifiers to 75 house groups,roughly half of which will be subscribers, served entirely by passive co-ax. With the analog co-ax runs pruned down to 75 or fewer houses, the amplifiers can be removed and the cable network can communicate in both directions at the same time on the same frequencies. Initially, this will allow US providers to try for 1 Gbps each way with FDX-capable CPEs. With upgrades this could reach over 5 Gbps capacity in each direction.
“This gets quite demanding for the cable infrastructure like the RemotePHY to manage,” Kurtz observes. “You can’t just pass the FDX RF signal through to the set-top box or cable modem. The large upstream signal from the guy next door will saturate the analog front end of your set-top box and knock out your service. So the RemotePHY has to to know which cable modems will interfere with which houses and carefully orchestrate who gets to talk and when.”
Moving the PHY out to the optical node also makes way for the next step: head-end virtualization. Today the head-end blockhouses are full of purpose-built hardware from specialized vendors. Tomorrow, with the MACs and cable PHYs gone, they can be full of commodity servers, running software to execute switching, network management, and control-plane functions. Some functions now performed in hardware by the head-end cable modem termination system (CMTS)—functions like the MAC—may require hardware accelerators on the servers. None the less, the network functions will move to software, Kurtz says.
Virtualization opens the door to even more change. Not only can the operator virtualize the core converged cable access platform functions—video, Internet, and voice—but they can bring in new services by simply running new applications code on the servers: content, games, retail, home automation—you name it. There is also the opportunity to pull many of the functions of the CPE back into the servers, reducing hardware costs and, incidentally, giving the service provider much greater visibility of how customers are using the network.
The cable system at this point would look entirely different than it does today. Gone is the big-iron head end, replaced by a miniature data center. Gone is the analog fiber, replaced, in all likelihood, by an inexpensive passive optical network (PON) offering symmetric up/down data rates. Gone is much of the complexity from the customer’s little boxes.
And that PON may start creeping closer to your house, as the fiber to the curb followed by 50-100 feet of remaining co-ax becomes an attractive approach. Or the last mile could become Ethernet. Or, since our series of steps has taken us near to 2020, the last mile in difficult terrain could be wireless.
Speaking of wireless, this new converged network could also serve as a backhaul medium for small cells, putting the most forward-looking operators in a position to move into the 5G wireless service business. That one physical resource—a few hundred feet of buried co-ax ending in your wall, so primitive but so hard to duplicate—makes the cable service providers incumbents to the future of ubiquitous connectivity. What they do with their incumbency is up to them—and to natural selection.
For Further Reading:
Explore FPGA applications in video distribution.