Why fiber must power the superfast broadband futureWhy fiber must power the superfast broadband future
by Nate Anderson | Published: April 10, 2008 - 10:45PM CT
Note to future self
Imagine your Future Self, going about your disease-free, chemically-enhanced life in 2015. Yes, you have a jetpack and yes, you own a robot, but the key question is how much bandwidth will you need to watch your IPTV, browse your favorite Web 4.0 sites, download your (legal) P2P content, and video conference with your robotophobic mother-in-law? And will it be available?
A new report (PDF) out from the Organization for Economic Cooperation and Development (OECD) suggests that your Future Self will need a 50Mbps download link in the next decade, and it surveys the worldwide race to get there that's raging between fiber to the home, VDSL, and DOCSIS 3.0 cable systems. In one important sense, though, fiber has already won.
Fiber triumphant
Despite all the hype about wireless technologies such as LTE and WiMAX, Rudolf van der Berg of the Netherlands' Ministry of Economic Affairs argues that none will be sufficient to provide widespread access to 50Mbps connections in the period between 2010 and 2020. Wired connections will continue to pummel their nerdier wireless siblings when it comes to sustained data transfer, which means that "wireless will be used to bridge the first meter, but not the first mile."
That means that the race to 50Mbps connections will largely be won by fiber-based networks; the only question is how close to the home those fiber lines run, and what sort of connection they use to get the signal the rest of the way. While fiber, DSL, and cable are generally considered separate technologies, most modern networks already incorporate fiber.
Cable plant in the US is generally a hybrid fiber-coax (HFC) system that runs fiber out to local nodes, where signals are converted to electricity and run to neighborhood homes over coax connections. DSL is moving the same way, as we can see in AT&T's current U-Verse deployment in the US; fiber is pulled to the local node, which is especially crucial for a distance-limited technology like DSL. And fiber run all the way to the home, like Verizon's FiOS and similar deployments in Japan, Korea, and France, is (obviously) all about the fiber.
Cable
All three of these approaches can hit the 50Mbps threshold... with some work. Cable requires an upgrade to DOCSIS 3.0, a system that uses channel bonding (sending the download signal over more than one empty TV frequency slot on the coax cable in order to boost capacity) to deliver speeds that could top out at 160Mbps in future. Comcast is deploying DOCSIS 3.0 this year, and has already turned it on in Minneapolis, where users can now get 50Mbps Internet connections for the low, low price of $150 a month.
The truly excellent news for cable is that coax can carry plenty of data, far more than traditional phone copper wiring. When (or if) cable companies eventually make the move to a fully IP-based network and adopt IPTV to deliver video, all of the bandwidth currently used to deliver analog and digital video to subscribers can be re-allocated to Internet use. Using even current technology, a cable line could carry up to 5Gbps under such a scenario.
DSL
Current (ADSL2+) DSL rollouts generally top out around 25Mbps, but a newer version called VDSL can reach up to 50Mbps and could keep DSL technology competitive for quite a while. This is especially important to companies like AT&T with huge existing investments in DSL infrastructure; while the company may well be forced to extend its fiber pulls all the way to the home in the next decade or two, Verizon's own examples show that the costs of doing so can be astronomical ($18 billion-$20 billion allocated so far). VDSL's problem, though, is that these high speeds tail off quickly with distance. At only 1km from the central switching office, the maximum speed has already declined to 30Mbps.
Few homes are located that close to the DSLAMs in the central office; van der Berg puts the number at less than 10 percent of the population in most countries. To deliver the full 50Mbps experience, VDSL connections should be less than 450 meters long, which means that a hybrid fiber system is absolutely essential to see these kinds of speeds.
To get a sense of just how expensive even such a limited fiber rollout can be, consider than telecom provider KPN in the Netherlands will expand from 1,350 local exchanges to around 28,000 in order to launch VDSL service. That's a lot of trench digging and cable laying to bring the fiber further into the neighborhood, making this a nontrivial upgrade.
But van der Berg clearly sees fiber to the home as the endgame. With its almost infinitely-expandable capacity (if you want more bandwidth, you can light up a new wavelength without pulling new fiber), a purely fiber-based system provides enough headroom for the foreseeable future.
Back in 1977, when the first commercial fiber optic connection was made in Long Beach, California, it carried only 6Mbps of traffic and required two fibers, one for each direction. Now, 3.2Tbps are possible per fiber, a million-fold increase in bandwidth. Van der Berg notes that lab tests using more than 160 separate wavelengths per fiber already allow for speeds up to 25Tbps.
Clearly, your Future Self wants this in the home. But laying a new infrastructure that stretches all the way to each individual doorbell is hugely expensive. The good news is that fiber bundles are light, weighing around 16 times less per kilometer than bundles of twisted pair wiring. They cost about the same to manufacture, too, and fiber is not susceptible to shorts from groundwater or electromagnetic interference. Added bonus: "rats and ferrets have been taught to pull fibre in situations where normal technologies do not work." (Seriously. Check page 22.)
The end of unbundling?
So your Future Self has fiber, probably with far more than 50Mbps of bandwidth, and your Future Self probably starts to think that he/she/it is living the digital dream. But there's a hitch. By moving the "central office" that exists in both cable and telephone-based systems out into thousands of smaller "central offices" at each local node, even a hybrid fiber infrastructure could make it much harder to share lines.
Local loop unbundling, as this is called, is common in Europe and helps to create significant competition among ISPs who would not otherwise be able to build an entire physical infrastructure on their own. This could be much harder under many future network scenarios. "Local loop unbundling," says the report, "may not be effective under certain network configurations. This could mean that incumbents may regain market power or, at best, if cable is available a duopoly situation may emerge."
Surprise! That's basically how things have turned out in the US after the FCC endorsed a "facilities-based" approach to competition (between DSL and cable, for instance) and got rid of mandatory line-sharing rules for DSL.
Even if line-sharing is still desired by regulators, it may be more difficult in practice. It turns out to be much cheaper for alternative ISPs to provide one backhaul link from a central office serving 70,000 people to their own network. Doing that from 20 local nodes is substantially more expensive and could also prove technically difficult on systems that share bandwidth, such as cable and fiber PON networks (separating out one user's traffic is much harder than on point-to-point systems like DSL).
"For alternative operators who use the unbundled local loop," writes van der Berg, "the business case is often not positive and they will either have to take the next step up the ladder of investment or will need to take a step back to offering wholesale broadband access."
Your Future Self might find that, without regulation, you may have few alternatives at all. Some users are already started to wonder about this after subscribing to FiOS (which generally gets quite high marks), as Verizon can remove the old copper infrastructure during the install (see here, here, here, and here). While the old telephone network was regulated to allow users to subscribe to the local and long distance phone provider of their choice, a Verizon-installed fiber line doesn't currently appear to permit such switching.
These are the kind of issues that will need to be hashed out as we continue our collective march into a glorious broadband future. As van der Berg makes clear, though, that future is bound to be based on fiber, even if copper and coax wires continue to enter our homes.
Blogged with the Flock Browser
