Stanford EE Computer Systems Colloquium

4:15PM, Wednesday, October 18, 2006
NEC Auditorium, Gates Computer Science Building B03

Measurements vs. Bits: Compressed Sensing meets Information Theory

Dror Baron
Rice University
About the talk:

Sensors, signal processing hardware, and algorithms are under increasing pressure to accommodate ever larger data sets; ever faster sampling and processing rates; ever lower power consumption; and radically new sensing modalities. Fortunately, over the past few decades, there have been enormous increases in computational power. This progress has motivated Compressed Sensing (CS), an emerging field based on the revelation that optimization routines can reconstruct a sparse signal from a small number of linear projections of the signal. The implications of CS are promising for many applications and enable the design of new kinds of cameras and analog-to-digital converters.

Information theory has numerous rich insights to offer CS. We investigate three directions along the interface between these fields. First, we characterize the minimum number of measurements needed to reconstruct the signal within a specified distortion - a problem related to rate distortion theory. Our study reveals that the unavoidable noise in analog measurements is the crucial factor that dictates the number of CS measurements needed. Second, we leverage the remarkable success of LDPC channel codes to design low-complexity CS measurement and reconstruction algorithms. Third, our work on distributed compressed sensing (DCS) provides new distributed signal acquisition algorithms that exploit both intra- and inter-signal correlation structures in multi-signal ensembles. We describe three models for signal ensembles, propose algorithms for joint recovery of multiple signals, and establish a parallel between the number of measurements needed and the Slepian-Wolf theorem from information theory. DCS is immediately applicable to a range of problems in sensor networks and arrays.

For more information, please view our CS resource page.


Download slides in Adobe PDF format.

About the speaker:

Dror Baron received the B.Sc. (summa cum laude) and M.Sc. degrees from the Technion - Israel Institute of Technology, in 1997 and 1999, and the Ph.D. degree from the University of Illinois at Urbana-Champaign in 2003, all in electrical engineering. From 1997 to 1999 he worked at Witcom Ltd. in modem design. From 1999 to 2003 he was a research assistant at the University of Illinois at Urbana-Champaign, where he was also a visiting assistant professor in 2003. Since 2003 he has been a postdoctoral research associate in the Department of Electrical and Computer Engineering at Rice University. His research interests include information theory, signal processing, and compressed sensing.

Contact information:

Dror Baron
drorb AT ece DOT rice DOT edu