Sign In

Communications of the ACM

ACM TechNews

Perfect Communication With Imperfect Chips


View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
Lav Varshney

MIT research affiliate Lav Varshney

Photo courtesy of MIT

Massachusetts Institute of Technology (MIT) research affiliate Lav Varshney has shown that some of the most commonly used codes in telecommunications can guarantee the reliable transmission of information even when the decoders are noisy. Varshney's analysis also shows that memory chips can preserve data indefinitely even when their circuits sometimes fail.

The key to the new analysis was not to attempt to quantify the performance of certain codes and decoders. Instead, Varshney looked at the statistical properties of whole classes of coders and decoders, showing that an average set of noisy decoders could ensure faithful reconstruction of corrupted data.

"It's not very intuitive to say that this kind of scheme can work," but Varshney's analysis draws conclusions by considering what happens as the length of the encoded messages approaches infinity, says Marvell Semiconductor's Shashi Chilappagari. However, he says chipmakers will be reluctant to adopt the coding scheme Varshney proposes without "time to test it and see how it works on a given-length code."

From MIT News
View Full Article

Abstracts Copyright © 2011 Information Inc. External Link, Bethesda, Maryland, USA 

 


 

No entries found