[Zurück]

M. Rupp:
"Robust Design of Adaptive Equalizers";
IEEE Transactions on Signal Processing, 60 (2012), 4; S. 1612 - 1626.

Although equalizers promise to improve the signalto-
noise energy ratio, zero forcing equalizers are derived classically
in a deterministic setting minimizing intersymbol interference,
while minimum mean square error (MMSE) equalizer solutions
are derived in a stochastic context based on quadratic Wiener
cost functions. In this paper, we show that it is possible-and
in our opinion even simpler-to derive the classical results in a
purely deterministic setup, interpreting both equalizer types as
least squares solutions. This, in turn, allows the introduction of a
simple linear reference model for equalizers, which supports the
exact derivation of a family of iterative and recursive algorithms
with robust behavior. The framework applies equally to multiuser
transmissions and multiple-input multiple-output (MIMO) channels.
A major contribution is that due to the reference approach
the adaptive equalizer problem can equivalently be treated as
an adaptive system identification problem for which very precise
statements are possible with respect to convergence, robustness
and l2-stability. Robust adaptive equalizers are much more desirable
as they guarantee a much stronger form of stability than
conventional in the mean square sense convergence. Even some
blind channel estimation schemes can now be included in the form
of recursive algorithms and treated under this general framework.

Although equalizers promise to improve the signalto-
noise energy ratio, zero forcing equalizers are derived classically
in a deterministic setting minimizing intersymbol interference,
while minimum mean square error (MMSE) equalizer solutions
are derived in a stochastic context based on quadratic Wiener
cost functions. In this paper, we show that it is possible-and
in our opinion even simpler-to derive the classical results in a
purely deterministic setup, interpreting both equalizer types as
least squares solutions. This, in turn, allows the introduction of a
simple linear reference model for equalizers, which supports the
exact derivation of a family of iterative and recursive algorithms
with robust behavior. The framework applies equally to multiuser
transmissions and multiple-input multiple-output (MIMO) channels.
A major contribution is that due to the reference approach
the adaptive equalizer problem can equivalently be treated as
an adaptive system identification problem for which very precise
statements are possible with respect to convergence, robustness
and l2-stability. Robust adaptive equalizers are much more desirable
as they guarantee a much stronger form of stability than
conventional in the mean square sense convergence. Even some
blind channel estimation schemes can now be included in the form
of recursive algorithms and treated under this general framework.

Blind channel estimation, convergence, iterative adaptive filters, least mean squares, linear equalizers, l2-stability, minimum mean square error (MMSE), recursive adaptive filters, reference modeling, robustness, zero forcing (ZF)

http://dx.doi.org/10.1109/TSP.2011.2180717

http://publik.tuwien.ac.at/files/PubDat_207068.pdf

Projektleitung Markus Rupp:
Signal and Information Processing in Science and Engineering II: Theory and Implementation of Distributed Algorithms