Lowly DSL poised for gigabit speed boost

DSL was one of the first widely adopted technologies for bringing high-speed Internet access to homes and businesses, but it hasn't been the fastest.

That's all changing. At the Broadband World Forum in Amsterdam this week, several companies are announcing and demonstrating products that bring DSL -- or digital subscriber line -- into a future with a speed of 1 gigabit per second. That's about 1,000 times the data-transfer speed the technology offered when it arrived in the late 1990s.

The DSL upgrade comes through a new technology called G.fast. Among those making network equipment chips to enable the technology are industry giant Broadcom, China-based Triductor Technology and Israeli startup Sckipio. The technology itself should arrive in homes starting in 2016.

The reasons people want fast broadband are plentiful: video chat with friends, high-definition TV from services like iTunes and Netflix, online backup of family photos, streaming music from Spotify, and an easier adjustment to the new era when apps are updated frequently. You can multiply all these uses by the growing number of devices in homes that tap into the Internet -- laptops, mobile phones, tablets, game consoles, thermostats, TVs, security systems. With a maxed-out gigabit connection, you can download a 4GB high-definition movie in about half a minute.

In many countries, though, particularly the United States, cable TV wiring has led the way for high-speed broadband. And when Google decided it wanted to accelerate the arrival of this high-tech future, it picked an even faster fiber-optic lines for 1Gbps today. That's all given a bad rap to DSL, which uses the phone systems' twisted pairs of copper wires that are more susceptible to radio-frequency interference.

With G.fast, Internet service providers and carriers will get a new way to give a new speed boost to DSL. And even though the upgrade will be expensive, it's necessary, said Ovum analyst Kamalini Ganguly.

"In many countries, doing nothing is not an option any more in my opinion," Ganguly said. "In particular, this is true where there is significant coverage and competition from cable companies, who I expect to be embarking on another upgrade over the next few years that will enable them to support 1Gbps services....We are also likely to see some fiber to the home."

DSL upgrades have at least one big advantage: they're an upgrade to networks that already exist. Much of the world doesn't have cable-TV infrastructure at all, and still less of it has fiber-optic connections. Phone networks, though, are widely used, and covered about 422 million DSL subscribers globally in 2013, according to analyst firm IHS.

That should rise to 480 million by 2018. But reflecting the competitive threat to DSL equipment makers, fiber optic links are expected to spread much more rapidly -- from 113 million in 2013 to 200 million in 2018.

Vectoring and G.fast

It will be awhile -- a couple years at least, most likely -- before DSL customers see gigabit speeds. The telecommunications industry really is only now installing its predecessor, called VDSL for "vectored" DSL. But the first stage of the next transition is beginning with the arrival of chips for the communications gear in telecom networks -- and in the home network equipment on the other end of the connection.

At the conference, Triductor will demonstrate its G.fast chip, which it plans to ship in the first quarter of 2015. Sckipio, too, plans G.fast demonstrations: its DP3000 chip for network gear and its CP1000 chip for the home network equipment.

Broadcom is pushing G.fast chips both for home gear and network infrastructure, showing off prototypes of its electronics.

The company won't say how much it is charging, but it's clear what the sales pitch is: a more affordable way for network operators to speed up broadband.

"We think it's very competitive relative to the alternative of deploying fiber," said Jim McKeon, a Broadcom senior director of product marketing. "We believe we are going to be the first to market with this. We're very excited for the potential for G.fast to unlock the hidden value of existing copper plant that is distributed worldwide for DSL."

As a standard, G.fast isn't quite done yet -- McKeon expects ratification at the end of the year or early in 2015 -- but firmware updates will let Broadcom tweak its products to comply with final refinements. The technology is governed by a pair of standards at the International Telecommunication Union -- ITU-T G.9700 and ITU-T G.9701.

Physics constraints

Faster broadband is great, but there's a good reason it's expensive for ISPs and telecommunications companies to build out their infrastructure: physics.

Broadband today works by transmitting electrical signals down wires. High-frequency signals whose voltage changes rapidly can carry more ones and zeros in a given amount of time. But the quality of high-frequency signals degrades faster than for lower frequencies, limiting the practical wiring length.

That's why broadband providers have been placing their network gear closer to homes -- often in boxes under sidewalks, in cabinets by roads, or boxes attached to telephone poles. That's also why it's so expensive to upgrade broadband networks: the ISPs have had to extend their networks to bring that network gear closer to their customers.

To meet its full gigabit-per-second potential, G.fast connections will require broadband providers to use network equipment close to the customers' buildings -- 50 meters (about 160 feet) or less, McKeon said. A 200-meter distance will still be good enough for about 600Mbps, he said. Some customers likely will bring their network gear all the way to a building, then use G.fast to provide connections to customers inside without having to replace the building's copper wiring.

G.fast requires more than just shorter distances for top speed. It also requires the vectoring technology introduced with VDSL. Vectoring is a technology that minimizes the interference between different customers' copper communication lines.

The basic problem is that copper wires act like broadcasting and receiving antennas, McKeon said. A signal in one line can cause a fainter signal in a nearby second line, muddying its signals. Meanwhile, the second line's signals interfere with the first. Multiply that by the number of lines in a network cabinet, and you've got a problem.

What vectoring does is take advantage of the fact that communications chips know what data will be sent over which lines and when. Complicated calculations can then be used to augment each signal's line with a different voltage to counteract the expected interference. That, in effect, tidies things up so high-frequency signals can be sent.

Gigabit push

Broadband providers are anxious to get high-speed connections to customers. "Being perceived as a technology leader was the overwhelming driver for offer gigabit broadband services," according to results published Monday in a Broadbandtrends survey (PDF) of 88 providers.

One early G.fast supporter is Telekom Austria, which last week announced it's got the technology working in real-world tests. Widespread installation is set for 2016.

"This technology will enable us to offer urban areas data rates 10, even up to 20, times higher than ever before," Telekom Austria CEO Hannes Ametsreiter said in a statement. "Fiber to the home remains our long-term vision, but we consider G.fast as an intelligent interim solution until fiber will have a similar coverage as we have with copper now."

European customers are likely to favor G.fast in particular, Triductor CEO Tan Yaolong said. That's because labor costs are very high in that region, which discourages extensive renovation projects.

Expense aside, fiber competition with DSL is very real. Google has increased its visibility with its Google Fiber project, which is providing 1Gbps broadband in Kansas City, Missouri, Kansas City, Kansas, and Provo, Utah, with Austin, Texas, on the way. US broadband giant AT&T has begun countering Google with its GigaPower program.

In the longer run, wireless broadband will play a role too. The US Federal Communication Commission last week announced (PDF) that it's examining wireless frequencies of 24GHz and higher: "It was long assumed that higher spectrum frequencies -- like those above 24 GHz -- could not support mobile services due to technological and practical limitations. New technologies are challenging that assumption and promise to facilitate next-generation mobile service -- what some call '5G' -- with the potential to dramatically increase wireless broadband speeds."

But wireless broadband suffers from capacity limits that aren't as much a burden for fixed-line communication technology like DSL, cable TV, and fiber optics.

"The demands of applications like multiple 4K TV streams to a household mean that serving the home via wireless isn't viable," said Ovum analyst Steven Hartley. "There are segments for which wireless makes sense -- low-end users, short-term renters, migrants. But these are also likely to be lower-spending segments."