EECS CSPL SEMINAR SERIES
WINTER TERM 2000
Dennis Goeckel
Electrical and Computer Engineering
University of Massachusetts - Amherst
goeckel@ecs.umass.edu
Thursday, March 2
4:00 - 5:00 P.M.
Room 1001 EECS
Adaptive Communication under Measurement Uncertainty
Abstract-
Because of the wide range of environments encountered by a wireless
communication system as conditions evolve over time, it is
well-established that performance gains can be obtained if
the
system is able to adapt to the current transmission
environment.
However, by definition, this adaptation must rely on
measurements of the
current environment; such measurements are inherently
outdated and/or
noisy, particularly in a system supporting high user
mobility. In this
talk, we will concentrate on adaptive system design in
environments where
there is a well-accepted class of statistical models that
characterize how
the current environment depends on available measurements.
In particular,
the problems of: (1) adaptive coding based on multipath
fading estimates,
and (2) adaptive power control based on estimates of the
log-normal
shadowing, will be considered in detail. The focus will be
on how the
uncertainty in the measurements alters the basic design
philosophies of
the adaptive system in each of these cases. Numerical
results will be
presented that demonstrate both the impact of considering
measurement
uncertainty in system adaptation and the general
benefits/detriments of
system adaptation. Issues of robustness to model
uncertainty will also
be addressed.
return to
CSPL
Seminars
homepage