This is every PC user's best dream, every bit – to have a computer that never crashes. New Scientist reports that there is now a self-repairing machine operating at the University College London (UCL) that can get back up after crashes, by repairing corrupted data. The computer reportedly works on the principle of nature, its “apparent randomness”.
Its self-repairing mechanism is such that it could keep systems that are important to missions working. Among the many areas that such a system could prove useful in, are instances wherein it can help drones to reprogram themselves to deal with combat damage or “help create more realistic models of the human brain”.
Self-repair? Really? (Image credit: Getty Images)
UCL computer scientist Peter Bentley, as per this report, is of the opinion that computers should be able to mimic nature's processes. “Its processes are distributed, decentralised and probabilistic. And they are fault tolerant, able to heal themselves. A computer should be able to do that,” he says. Shedding some light on what we are looking at with this new system, the report adds that in this system, data is tied with instructions on what to do if. Say, it links the outside temperature with instructions on what to do if the temperature gets unbearable. “It then divides the results up into pools of digital entities called 'systems',” the report adds.
Speaking on the kind of systems we work with, the report adds that these work on a set of instructions. These are done over and over again, controlled by a sequential timer called program counter. Bentley believes that while this way a system may be able to do the necessary, it “doesn't lend itself to simultaneous operations”. He is further quoted as saying, “Even when it feels like your computer is running all your software at the same time, it is just pretending to do that, flicking its attention very quickly between each program.”
Explaining further, he adds that every system has a memory with “context-sensitive” data, implying that it interacts with similar systems. Importantly, there is no program counter at work here; instead, the system executions are based on times picked by a number generator that are designed such that they are like the randomness in nature. These systems work simultaneously, with neither of them suppressing the other. Bentley adds, “The pool of systems interact in parallel, and randomly, and the result of a computation simply emerges from those interactions.”
Coming to its self-healing nature, the report elaborates that the set of instructions is distributed across several of its systems. This way, if one system gets corrupted, then the computer can access another uncorrupted copy to carry out the repair. What's interesting here is that unlike regular systems that crash when they are unable to access a bit of memory, this system will continue functioning, since each system has its own memory.