Communication Theory in the
Cloud
A Joint Event with the Toronto Chapter of the IEEE
Communications Society
Matthew C. Valenti
Lane Department of Computer Science and
Electrical Engineering
West Virginia University
Friday, October 30, 3pm
Location: BA 1210
Abstract:
With the advent of Cloud
Computing, it is now feasible for even small research groups to apply massive
computing power to the solution of hard problems. For instance, accurate Monte
Carlo simulations may be rapidly executed in parallel using grid computing
technology, and systems with large design spaces may be efficiently optimized
with evolutionary computing techniques.
Under funding by the National Science Foundation, West Virginia University is
developing an open infrastructure for the execution of communication-theory
related tasks on a massive public grid computer. When completed, the project
will feature a web-based interface that will allow researchers from around the
world to easily run jobs on a computing cloud that we maintain. Casual users
can run jobs without needing to write a single line of code by using a Matlab-based
library developed at WVU called the Coded-Modulation Library (CML), while
authorized expert users may run their own custom code. Rather than relying on
dedicated computing resources, the project is based on the idea of community
computing, whereby participants in the project contribute idle cycles on their
ordinary desktop computers by running a low-priority background process (for
linux) or screensaver (for windows).
In this talk, an overview of our Cloud Computing infrastructure will be
presented. We will discuss an approach for creating a community-computing grid
and a methodology for running code written in Matlab on the grid. In addition,
several case studies will be presented showing how the computing grid may be
used to optimize communication waveforms. In particular, we show how massive
simulation and genetic algorithms may be used to optimize: (1) coded
continuous-phase modulation (CPM) waveforms, (2) label mapping and signal
design for bit-interleaved coded modulation (BICM), and (3) dispersion
matrices for space-time block codes (STBCs).
Bio:
Matthew
C. Valenti has been with West Virginia University since 1999, where he is
currently an Associate Professor in the Lane Department of Computer Science and
Electrical Engineering. He holds BS and Ph.D. degrees in Electrical Engineering
from Virginia Tech and a MS in Electrical Engineering from the Johns Hopkins
University. From 1992 to 1995 he was an electronics engineer at the US Naval
Research Laboratory. He serves as an associate editor for IEEE Transactions on
Wireless Communications, was a co-chair for the Wireless Communications
Symposium at ICC-2009 (Dresden, Germany), and is a co-chair for the
Communication Theory Symposium at ICC-2011 (Kyoto, Japan). His research
interests are in the areas of communication theory, error correction coding,
applied information theory, wireless networks, simulation, genetic algorithms,
and grid computing. His research has been funded by the National Research
Foundation and the Department of Defense.
Host of Talk:
Raviraj Adve (rsadve@comm.utoronto.ca)