Accepted to IEEE Transactions on Vehicular Technology, 2004.
In this paper, a two stage hybrid interference
cancellation and equalization framework is proposed for interference
cancellation in the uplink of multiple-input multiple-output
orthogonal frequency division multiplexing (MIMO-OFDM) systems.
The first stage uses time domain equalization to suppress
co-channel interference, mitigate asynchronism, and shorten the
post-equalization channel response to be no longer than the
length of the cyclic prefix. The second stage performs lowcomplexity
single tap equalization and detection in the frequency
domain. The framework is developed specifically for spatial
multiplexing and is applied to multiuser MIMO-OFDM systems
with asynchronism between users as well as to single-user MIMOOFDM
systems. Various equalizer design methods are proposed
that determine the coefficients directly from the training data
and are compared with methods based on channel estimates. The
equalizer coefficients and post-equalization channel response are
found by solving a joint optimization that maximizes the signal
to interference-plus-noise ratio (SINR) in the frequency domain.
Simulations compare various training based methods and show
the proposed methods provide good bit error rate (BER) and
SINR performance in a variety of interference scenarios.
MIMO, OFDM, interference, cancellation, space-time.
This paper is available in .pdf format.