Title:
Superfast algorithms for digital resampling between uniform and non-uniform grids

Abstract:
Resampling discrete signals is a fundamental task in digital signal processing. For many DSP applications (packet data traffic, geophysics, medical imaging, and astronomy, for example), resampling is used to convert data between uniform and non-uniform grids, which eases both the acquisition process and the data analysis.  This talk presents new resampling algorithms that use the Fast Fourier Transform (FFT) and Fast Multipole Method (FMM) to resample between uniform and non-uniform sampling grids through sinc interpolation. These algorithms are based on several results from structured linear algebra, and yield highly accurate resampling with FFT-like asymptotic complexity.  We demonstrate the potential utility of these algorithms by considering the computational savings they could offer to level-crossing ADCs.

Speaker Bio:

Chris Turnes received his B.S. in electrical engeering from the University of Maryland College Park in 2007 and his M.S. from Georgia Tech in 2010.  He is currently a Ph.D. candidate in electrical engineering at Georgia Tech, under the direction of Justin Romberg.  His research interests are efficient algorithms for structured linear algebra and their applications to signal processing.

Chris leads an interesting personal life, as well.  In high school, his superlative was given as simply "Most."  He is the co-creator of NBC's "Grads," and is credited as one of the authors of the original Port Huron statement (not the compromised second draft).  In a tribute to his musical prowess, he was once chosen to host the Grammys, giving every award to Cory Hart.  He enjoys reading, and has many leather-bound books that he reads in his condo (which smells of rich mahogany).  He is said to be the life of parties he's never attended, and it is rumored that the film "The Bourne Ultimatum" is loosely based on his life.