Tag Archives: bell labs

Bell Labs test shows faster speeds on shorter copper

by Steve Blum • , ,

Next generation cable technology – DOCSIS 3.1 – can support symmetrical 10 Gbps speeds over hybrid fiber coax plant, according to a press release from Bell Labs, now known as Nokia Bell Labs. Nokia completed its purchase of Alcatel Lucent earlier this year and Bell Labs was part of the bargain.

Bell Labs is pitching its XG-Cable technology for integration into CableLabs’ DOCSIS 3.1 standard, which is undergoing field trials in a few U.S. markets. It’s essentially the same pitch that companies with G.fast gear are making to telcos: our stuff will dramatically boost broadband speeds on existing copper wire networks.

From everything I’ve seen, that’s true. At least as far as it goes. And that’s the catch. Both XG-Cable and G.fast are relatively short range technologies. Bell Labs’ says it simultaneously pushed 10 Gbps in both directions over 100 meters of coaxial cable on 1.2 GHz of bandwidth under laboratory conditions. The speed dropped to 7.5 Gbps symmetrical when a point-to-multipoint architecture was used. Ultimately, Bell Labs expects to get that level of performance at up to 200 meters.

G.fast, which is touted as an upgrade path for DSL networks, is limited to something like 250 meters to 500 meters of copper wire, with speeds ranging from 100 Mbps to 1 Gbps, again depending on how far it has to go (and who’s press release you’re reading). To get to the high end of the range, the distance needs to be less than 100 meters.

For either G.fast or XG-Cable to deliver promised speeds, the outside plant needs to be in good condition. Deteriorating lines will mean sharp drops in performance, to the point that older technologies will outperform it. With all due regard to the danger of taking an analogy too far, it’s not unlike trying to drive a Lamborghini on a dirt road.

But if you’re on a short, pristine track, a Lamborghini will fly.

It’s easy jump to the conclusion that technological advances such as these will render fiber unnecessary. That’s not true. The way you get short, final copper runs is to push more and more fiber, deeper and deeper into the network. At some point, it might not be necessary to go all the way to a home or business to get fiber-class broadband speeds, but you’ll have to extend the fiber portion of last mile networks much closer. And you’ll have to add fiber capacity – either more strands or better electronics – to handle the increased demand for bandwidth.

It’s still early days for this technology, but it’s good news that it might not be too far over the horizon.

Forget gigabits, Bell Labs says petabits are coming

by Steve Blum • , ,

What’s a petabit for, if not more grumpy cats?

The theoretical fiber speed limit continues to increase. Bell Labs says it’s successfully tested technology that has the potential for moving data through an optical fiber at the speed of 1 petabit per second. A petabit is 1,000 terabits, which in turn is 1,000 gigabits. Currently, the top speed for optical fiber is in the 10 terabit to 20 terabit range, according to Bell Labs. The technique involves multiplexing six separate transmissions on a single fiber

Using the MIMO-SDM technique, Bell Labs aims to overcome the non-linear Shannon limit of currently deployed optical fiber. During the 6×6 real-time MIMO transmission technology experiment, crosstalk from multiple signals on a special fiber supporting six parallel optical signal paths was removed for the first time using real-time processing. This breakthrough brings the technology a step closer to reality compared to previous experiements using off-line processing.

The MIMO-SDM technique has the potential to increase current fiber capacities to a Petabit-per-second — enough capacity to allow two-thirds of the U.S. population to simultaneously stream HD movies over a single optical fiber.

It’s a long way from the lab to commercial use, though. Bell Labs isn’t saying when petabit-class technology will actually be deployable, but its press release talks about it in the context of 5G mobile networks, which means it’s something like five to ten years out.

Which is fine. Today’s fiber capacity bottlenecks are, for the most part, due to continued use of cheaper, older equipment rather than the limits of the more expensive new stuff that’s on the market. Bell Labs’ breakthrough means that we can expect today’s expensive gear to become tomorrow’s bargain. So long as dollars drive network upgrades, meeting ever growing demand remains an economic choice rather than a physical technology limit.

Bell Labs goes looking for lost mojo

by Steve Blum • , , ,

First we’ll invent Unix, then we’ll figure out what to do with it.

When Silicon Valley was just pear orchards and a junior university, and a google was an obscure bit of math trivia, the wellspring of geek creativity was a continent away. Bell Labs sprawled across several campuses in northern New Jersey, filled with scientists and engineers who were paid to come up with interesting ideas and novel technology. Not necessarily marketable products, although it was correctly assumed that profits would follow somehow. Partly because the Bell System was a monopoly with profits more or less determined by regulators on a cost-plus basis, but mostly because Bell Labs delivered: the transistor, Unix, C, fiber optics, CDMA, TDMA and the list goes on.

The break up of AT&T thirty years ago brought an end to that cloistered world. Even as it was collecting Nobel prizes for past glories, Bell Labs was bouncing from one corporate parent to another, continually shedding talent and narrowing its scope, and finally ending up as a vestigal stump in an R&D backwater of Alcatel-Lucent.

It seems, though, that Alcatel-Lucent has woken up to the fact that the smart boys and girls are not just hanging out at the Jersey shore these days. Corporate CTO Marcus Weldon was also named president of Bell Labs in November and, according to LightReading, he’s going to try to lure them back…

Weldon will be looking to add some people to the Bell Labs team, though some will also be lost as part of [previously announced] layoffs. The R&D operation lost people to the web giants in Silicon Valley a decade ago. “We need to hire some people who would otherwise work there. We’ll be hiring some talent in that area.”

Weldon is promising to give Bell Labs interesting and important problems to solve. If he and his bosses have the wisdom and patience to let the answers fall naturally from those problems and not be driven by predetermined corporate roadmaps, some of the mojo might return.

I had a summer job at Bell Labs’ Piscataway campus in the 70s, but all I figured out was how to use this odd interconnecting network to log onto the system back in Berkeley and play Star Trek.

Bell Labs bridges a gigabit over a copper gap

by Steve Blum • , , , ,

The case for copper. Source: Alcatel-Lucent.

AT&T and Verizon should think twice about running away from older copper networks. Bell Labs has prototype technology that can already move half a gigabit through legacy wiring. Testing by parent company Alcatel-Lucent and Telekom Austria succeeded in pushing more half a gigabit over multiple legacy copper POTS pairs, using elements of the emerging G.fast standard and mixing in advanced vectoring technology – dubbed Vectoring 2.0 – developed by Bell Labs.

G.fast is an International Telecommunications Union standards initiative that’s intended to eventually develop technology that will allow 1 Gbps data speeds over copper telephone wiring at distances up to 250 meters.

There’s no free lunch. Although Bell Labs managed to hit 1.1 Gbps over 70 meters of a single, good quality copper pair and 800 Mbps at 100 meters with current G.fast gear, speeds dropped dramatically when older plant was tested. By itself, a legacy unshielded pair was able to carry 500 Mbps for 100 meters, but the crosstalk generated by lighting up a second pair in the bundle dropped throughput to 60 Mbps, putting it back into current VDSL2 territory.

Adding the Vectoring 2.0 technique cured the crosstalk problem. It expands the bandwidth of vectoring technology by more than a factor of ten, and brought throughput back up to 500 Mbps.

The objective is to create “the fixed networks equivalent of wireless small cells”. G.fast performance drops quickly as copper distances stretch out toward 200 meters or more. But if the promise holds and fiber is installed far enough into the network – think fiber to the curb or basement – homes might soon see gigabit speeds at significantly lower costs.