Distributed canonical correlation and distortion-rate analyses for centralized compression-estimation using wireless sensor networks

Thursday, June 30, 2005 - 2:30pm - 3:30pm
EE/CS 3-180
Georgios Giannakis (University of Minnesota, Twin Cities)
Wireless sensor networks deployed to perform surveillance and
monitoring tasks have to operate under stringent energy and
bandwidth limitations. These motivate well distributed compression
and estimation scenarios based on reduced dimensionality sensor
observations which may have to be severely quantized before
transmission to a fusion center. We will show how distributed
correlation analysis can be used to compress observations and
explore the fundamental performance limits dictated by distortion-rate
analysis in this decentralized estimation setup. We will further
present interesting tradeoffs that emerge even in distributed
mean-location estimation based on severely quantized observations
along with their fundamental error-variance limits. Estimators
utilizing either independent or colored binary data will be
developed and analyzed. Corroborating simulations will provide
comparisons with the clairvoyant estimators based on unquantized
sensor observations, and include a motivating application with a
sensor net employed for habitat monitoring. In the poster session
we will also discuss dynamical systems and present (extended)
Kalman Filtering ideas based on single-bit observations.

Brief Bio:
G. B. Giannakis received his B.Sc. in 1981 from the Ntl. Tech. Univ.
of Athens, Greece and his M.Sc. and Ph.D. in Electrical Engineering
in 1983 and 1986 from the Univ. of Southern California. Since 1999
he has been a professor with the Department of Electrical and
Computer Engineering at the University of Minnesota, where he
now holds an Endowed ADC Chair in Wireless Telecommunications.
His general interests span the areas of communications and signal
processing, estimation and detection theory -- subjects on which
he has published more than 200 journal papers, 350 conference papers,
and two edited books. Current research focuses on complex-field and
space-time coding, multicarrier, ultra-wide band wireless communication
systems, cross-layer designs and wireless sensor networks. He is
the (co-) recipient of six best paper awards from the IEEE Signal
Processing (SP) and Communications Societies (1992, 1998, 2000, 2001,
2003, 2004) and also received the SP Society's Technical Achievement
Award in 2000. He is an IEEE Fellow since 1997 and has served the
IEEE in various editorial and organizational posts.