Imagine transmitting the content of the entire Library of Congress in ten seconds. Yes, that’s fast. That communication speed translates to 26 terabits per second, which is, for now, the fastest speed attained by a communication system using a single laser beam and optical fiber.

Actually not so long ago people could barely imagine transmitting the Library of Congress in one go, in fact, the question might have been why would anyone want to? This does sound like a dumb question, or at least the kind of question you’d expect from somebody who has never used a computer on the Internet. However, another way to think about it – what would you do with the content of the Library of Congress if it arrived in your computer in ten seconds? (You’d better have a rather large rack of disk storage, for one thing.) So, okay for an individual this kind of speed isn’t relevant. Where it is relevant, of course, is in the ‘backbones’ (main transmission lines) of the Internet where millions of individual communications are happening every second, and where the sharp rise in video-on-demand is loading more data onto lines than ever before. In that context 26 terabits per second is good, not great but good.

Why not great? Let’s start with a couple of factoids to tuck into your memory cells: A terabit is 1,000,000,000,000 bits. That’s twelve zeros to the one. It makes for a large looking number. The other factoid is a phrase: orthogonal frequency division multiplexing, a technique that uses multiple lasers to create data streams in different colors and shoot them down the line together. This approach has reached 100 terabits per second. The problem is it takes 370 lasers, which are not cheap – nor is the energy requirement.

What the researchers at Kalrsruhe Institute of Technology (KIT), Germany have done is achieve the 26 terabit speed with one laser. Nature Photonics [22 May 2011, paywalled, 26 Tbit s?1 line-rate super-channel transmission utilizing all-optical fast Fourier transform processing]

The innovation is using a frequency comb to create exceedingly short pulses of colored light – up to 325 different colors at a time – and each color encoded with its own data stream. This very complex aggregate of data can be sent down an optical fiber cable and at the receiving end decoded using a mathematical routine known as a fast Fourier transform (FFT). Now here’s the real trick. A computer running the FFT could not process the incoming 26 terabit stream fast enough, so what the researchers did is implement the FFT in an optical solution. That is, they created a way to split the incoming beam using optics (essentially, lenses) into many paths, which creates a time lag between streams that is equivalent to how the FFT is calculated. These ‘calculated’ streams are then integrated on a single collector.

Therein lays a potentially important real-world advantage. Not only does this approach use only one laser and a lot less power, according to Wolfgang Freude, one of the lead researchers:

The current design outperforms earlier approaches simply by moving all the time delays further apart, and that it is a technology that could be integrated onto a silicon chip – making it a better candidate for scaling up to commercial use. He concedes that the idea is a complex one, but is convinced that it will come into its own as the demand for ever-higher data rates drives innovation.

[Source: BBC Science News]

Of course this is research, one setup transmitting a test message 50 kilometers. Making this work on thousands (or millions) of transmissions worldwide in reliability and cost factors necessary for commercial production – now there’s the real trick. But at a minimum this research shows that with ingenuity the need for speed (actually, high volume) can perhaps be met with lower costs and less energy, and that’s a worthy goal by any measurement.