Sunday October 15, 2017

Harnessing the power of discarded cell phones

Linking the processors of discarded cell phones together into a server could be an energy-efficient way of solving computational problems.

Linking the processors of discarded cell phones together into a server could be an energy-efficient way of solving computational problems.

Listen 7:10

If you're like many of us, the chances are, when you last upgraded your mobile phone, you just put the old one in a drawer somewhere. And it's gathering sock lint.

Or, worse, you threw it out, and it's now sitting in landfill somewhere leaching toxic metals into the ground.

But the thing is, the processor on that phone, the brain, so to speak, is probably still perfectly good. A little slow, maybe, but in working condition.

Most people replace their phones when the screen gets damaged, the battery stops working, or they can't properly charge it anymore, says Mohammed Shahrad, a PhD candidate at Princeton University in New Jersey.

Mohammad

Mohammad Shahrad (Princeton University)

But the circuit board on the phone, which contains the central processor, the graphics processor and various hardware accelerators, is usually fine, if a little slow.

So Mohammad came up with a great idea.

"I was in the shower," he says, when he began wondering whether he could take those old phones, divert them from the landfill (or sock drawers) and link them -- hundreds or thousands of them -- to form a cloud server.

"They're actually very powerful computers," he says.

So he designed a way to house groups of 84 cell phones into a cloud server, which could then be used to solve computational problems.

It's a win-win situation, he adds, noting that although there would be some energy required to run the servers, most of the carbon footprint created by mobile phones is in the manufacturing process.

"So by repurposing the phones, we're extending the footprint," he says, meaning that the net cost to the environment of each phone gets smaller the longer it's able to be used.

So far, it's just conceptual, he points out. But he and his colleagues hope to have a working prototype by the end of 2017.