Geneticists to the internet's rescue

Source: NICTA

Australian scientists have used genome analysis tools to create a patented technology to investigate the fate of the laser beams zapping through the optical fibres that connect our cities.

Their ideas have broken the back of a communications industry problem - how to identify the causes of noise in these optical cables that form a key part of the backbone of the internet.

The device that they and their fellow engineers at NICTA, Australia's Information and Communications Technology (ICT) Research Centre of Excellence, have invented, will, for a few thousand dollars, do a job that today would cost $100,000 plus and would require multiple types of test equipment.

And it will allow phone companies to confidently increase the speed ratings on long haul optical fibres - from 10 gigabits per second to 40 gigabits per second or more without losing data in the noise in line.

"We developed a way of presenting an optical signal as a two dimensional image," said NICTA principal researcher , Trevor Anderson, who is based at NICTA's Victoria Research Laboratory and directs the Managing and Monitoring the Internet (MAMI) Project.

"We thought that it would allow us to recognise the 'fingerprint' of the various kinds of optical noise that can interfere with the signal.

"But we didn't know how to analyse the image. Fortunately in the next door laboratory NICTA has a team of geneticists analysing vast lengths of genetic code to find patterns of gene sequences that would indicate a tumour.

"Dr Adam Kowalczyk looked at our problem and laughed - 'This is easy - biology is so much more complex,' he said to me.

"We have to identify cancer subtypes using a handful of noisy examples to learn from rather than the thousands that are available to you, so Dr Kowalczyk said let's try our algorithms on your data."

The result is a new device known as a multi-impairment monitor. And it can identify the distinct visual patterns created by the common forms of noise and distortion in optical fibres.

"The current tools available in the marketplace only count the errors in the data, telling the operator a problem exists but not what that problem is, where the problem is or what caused it," Anderson said. "Our device can already identify the top four sources of noise and we expect to be able to do all six."

He anticipates the device will be ready for market in 12 months.