AN IMMUNE system that can see off potentially fatal threats may soon be as important to electronic circuits as it is in our bodies. That鈥檚 the goal of immunotronics, part of a grand plan to use biological principles to create computer hardware that can repair or evolve new functional parts when needed.
Just as biological immune systems hunt down invading pathogens, immunotronics will be used to locate faults in electronic devices, and then repair them, says Andrew Tyrrell at the University of York. This should be especially useful for devices in harsh environments such as car engines where chips are most likely to break down, or for circuits out of reach of a technician such as those aboard a distant space probe or deep inside a nuclear reactor.
In your body, helper T cells identify foreign threats because of the unfamiliar proteins they produce, called 鈥渁ntigens鈥�. When T cells detect an antigen, they send a message to B cells, which produce antibodies that target the antigen.
Advertisement
In Tyrrell鈥檚 immunotronic system, the aim is to detect and repair errors-antigens-caused by a fault in any of the thousands of transistors built into a computer chip. The role of the T cells is to highlight errors by training them to recognise valid patterns of proper transistor operation.
During operation, a separate system analogous to antibody-producing B cells continually generates all the different patterns that the chip is capable of producing-whether valid or invalid. By comparing these patterns with those actually produced by the chip, it is possible to spot errors-and tag them for destruction, like antibodies do.
To ensure that the B cells don鈥檛 produce antibodies that match valid operations-and so turn on their host-they are first compared with those valid patterns already learned by the T cells. Just as in biology, any that match are ordered to self-destruct before they can do any damage by being released into the system.
Once an error is located, programmable chips called field programmable gate arrays are called into play. The system simply configures the FPGA to compensate for the fault. Tyrrell鈥檚 long-term goal, as part of a European Commission鈥檚 Future and Emerging Technologies project, is to take this a step further and have hardware breed new functional components using spare FPGA chips.
Some error detection systems, such as those used on the space shuttle, rely on voting schemes, where several computers vote on an outcome. If all work properly, they should agree. But if several of them fail, the result is going to be wrong, says Tyrrell.
Automatic fault correction is particularly important for remote systems. One of NASA鈥檚 aims is to send probes on hundred-year missions. 鈥淚n a hundred years a lot of things can go wrong,鈥� says Tyrrell.
