In January leading microelectronics academics, the UK Knowledge Transfer Network (KTN), the Engineering and Physical Sciences Research Council (ESPRC) and industry will meet to decide the future of what they call the grand challenges in microelectronics design.

These are: a neural network-based computer; a pocketable super-computer; a battery-less mobile phone; bio-silicon interfaces; and practicable alternatives to IC scaling. Each project requires funding of between £1m and £2m.

“There is no promise of any resources,” Steve Furber, Professor of computer engineering at Manchester University, told Electronics Weekly.

“The idea is that we, the UK microelectronics community, come up with something, some key ideas, and what happens next has to be decided,” he said.

The neural network brain is already being built at Manchester University using 20 ARM processors simulating about 1,000 neurons each.

Furber reckons they have found a way of getting past the inability of parallel processing systems to scale performance upwards in line with the number of processors. “We can see how to build a system that will scale up as far as we can see,” said Furber.

Another challenge under development is the battery-less mobile phone. Southampton University has had a project running called Vibes (vibration energy scavenger) which uses the vibrations incurred by carrying around the phone as an energy source.

The challenges came out of a feeling that the research community was “fragmented”, Professor Mark Zwolinski of Southampton University, told EW.

Zwolinski’s project is to develop techniques, such as redundancy and fault tolerance, to deal with very fine geometry processing where transistors and materials are likely to be defective.

Cracking the grand challenges is some way off. “It is not that the research will produce immediate, tangible benefits, but that the journey could be interesting,” said Zwolinski.