A team of British researchers is working to create commercially viable broadband speeds up to 2,000 times faster than the slickest of connections available today, and say that their project could become a reality in just three years.

It might sound like a fantasy, but the scientists at Bangor University in Wales are deadly serious about what they’re doing – and what’s more, they say that it won’t even require any new infrastructure to be built.

The BBC reports how the scientists have looked at other ways of increasing broadband speeds – which include adding more fiber optic strands into cables, using more lasers to code and decode data, and signal amplification technologies – and dismissed them as being far too expensive.

Instead of trying to build something new, the Ocean Project researchers are simply tweaking an existing technology known as Optical Orthogonal Frequency Division Multiplexing (OOFDM) that is already widely used in wireless networks to convert raw data into electrical waves, which can then be transmitted through fiber optic cables. What they’ve done is to develop an electronic system that can encode and decode optical signals on the fly, effectively turbo charging them up to hitherto unprecedented speeds.

So far, the team has hit speeds of a whopping 20Gbps – compare this to the current average download speeds in the UK right now, which stand at a mere 33.4Mbps according to ISPreview. Even the supposedly ‘super-fast’ Google Fiber project, currently being trialed in Kansas, can only deliver a 1Gbps connection.

Dr Roger Giddings, lead researcher at the Ocean Project, told the BBC that two challenges remain. The first is to get broadband speeds up to a lighting fast 40Gbps – a target that they consider to be very achievable in the next three years – and second, to follow that by coming up with a way to integrate their new system into existing fiber optic networks.

With the likes of Fujitsu Semiconductors Europe, Finisar Israel, and VPIsystems all onboard and providing expert assistance, they might just be onto something.