University of Arizona professor Bane Vasic has led the development of an error-correction decoder that outperforms belief propagation algorithms. Vasic compares the decoder to solving a Sudoku puzzle, in which the puzzle's cells represent transmitted data bits that need to be reconstructed.
"A neat property of this new algorithm is that there is no need for a brain or some central intelligence to solve the puzzle, because the cells solve the puzzle collectively," he says. "No individual cell has a global knowledge about the solution, but collectively the cells find the solution by passing messages among one another."
Originally, it was thought that the new algorithms would need to arrive at a solution that nearly solves the puzzle, but that would have made them too complex. Instead, the researchers found that in many cases the "simple decoder outperforms belief propagation to the extent that errors are reduced by 90 percent," says Vasic, who adds that the discovery "opens up a plethora of beautiful theoretical problems."
From UA News
View Full Article
Abstracts Copyright © 2010 Information Inc., Bethesda, Maryland, USA
No entries found