Systems Seminar
On the Scaling Laws of Dense Wireless Sensor Networks
Prof. Hesham El Gamal
The Electrical Engineering Department
The Ohio State University
Abstract
In this talk, we consider dense wireless sensor networks deployed to observe
arbitrary random fields. The requirement is to reconstruct an estimate of the
random field at a certain collector node. This creates a many-to-one data
gathering wireless channel. We first characterize the transport capacity
of many-to-one dense wireless networks subject to a constraint on the total
average power. We then use this result along with some information theoretic
tools to derive sufficient and necessary conditions that characterize the
set of observable random fields by dense sensor networks. In particular,
for random fields that can be modelled as discrete random sequences,
we derive a certain form of source/channel coding separation Theorem.
We further show that one can achieve any desired non-zero mean square
estimation error for continuous, Gaussian, and spatially bandlimited
fields through a scheme composed of single dimensional quantization,
distributed Slepian-Wolf source coding, and a channel coding strategy
optimized for the AWGN channel. Based on our results, we revisit earlier
conclusions about the feasibility of dense sensor networks. As argued
in the sequel, our results may have some important implications on the
design of communication protocols for dense sensor networks.
Time and Place: Tue., Apr. 15, at NOON in 4610 Engr. Hall.
*** NOTE SPECIAL DAY & TIME ***
SYSTEMS SEMINAR WEB PAGE:
http://www.cae.wisc.edu/~gubner/seminar/