Nanomaterial May Allow Devices to Rewire Themselves 61
An anonymous reader writes "Researchers have developed a nanomaterial that can 'steer' electrical currents. The discovery could lead to the invention of devices that can reconfigure their internal wiring and evolve into an entirely different and new device, to reflect the changing needs of consumers. From the article: 'The team is aiming to create a single device able to reconfigure itself into a resistor, a rectifier, a diode and a transistor based on signals from a computer. The multi-dimensional circuitry could be reconfigured into new electronic circuits using a varied input sequence of electrical pulses, the team said. 'Our new steering technology allows use to direct current flow through a piece of continuous material,' said Professor Bartosz Grzybowski, who led the research. 'Like redirecting a river, streams of electrons can be steered in multiple directions through a block of the material; even multiple streams flowing in opposing directions at the same time.'"
Old news (Score:5, Interesting)
The discovery could lead to the invention of devices that can reconfigure their internal wiring and evolve into an entirely different and new device, to reflect the changing needs of consumers
No that's old old old stuff not an invention.
In the digital world, think of a classic digital computer. Decade(s) ago I've used (expensive telco) FPGA products which reconfigure themselves. Some of the exotic massively redundant switchgear could reconfigure itself on the fly while passing production customer traffic, although we usually did it during maintenance windows anyway. VLIW CPUs, etc. I've done embedded FPGA work where you embed a really simple CPU in the FPGA and build all the smarts into the FPGA as reconfigurable peripherals of the CPU itself, so you start with a minimal but usable "microblaze" (or was it picoblaze?) core and then add a hardware multiplier as necessary, etc. Very old stuff, not new...
In analog you've got the option of doing it "for real" with analog computer building blocks and lots of analog switches, or doing it "emulated" using DSP chips.
This approach is currently economically feasible, but rarely implemented. Mostly terror of being single sourced, or violating a patent. If I buy a USB interface that violates someones patent, I'm much more insulated than if I implement a FPGA / software USB interface that violates that patent. Maybe not legally, but definitely practically.
It might be new in that its yet another implementation, kind of like "yet another ia32 386 compatible CPU" can be new. It might be new in that its really freaking small or really energy efficient (although existing DSP chips, shipping in the millions, are going to crush your R+D possibly beating a theoretically better technology)