34. Better Math Makes Faster Data Networks

Improved technique can process signals 1,000 times faster than at present.

By Gillian Conahan|Monday, January 21, 2013

Check out these numbers: Researchers at MIT have figured out an improved mathematical technique that can process signals up to 1,000 times faster than is possible with today’s technology. Their approach promises to turbocharge GPS, MRI scans, and many other data-intensive processes.

The advance improves on a staple of modern computation known as the fast Fourier transform, or FFT, which breaks down a complex signal into its component parts—almost like describing a piano chord by identifying the individual notes that need to be played. In January computer scientists Dina Katabi and Piotr Indyk, along with their students Eric Price and Haitham Hassanieh, announced a way to get the job done much faster.

Signals that have the fewest component frequencies, like those used for medical imaging, will see the biggest speed gains, but more complex tasks, such as video file compression, should get a substantial boost as well. “Being able to move on this problem that touches on so many fields excites us tremendously,” Katabi says. The benefits could arrive within five years.

Comment on this article