Low Complexity Turbo Space-Time Equalization for Broadband MIMO Systems

INTRODUCTION

In this paper, we consider turbo equalization formultiinput multi-output (MIMO) broadband wireless transmission that is aected by multipath fading. Note that, due to fading eects at high data rates, the wireless link between each transmit/receive antenna pair is represented by a long channel impulse response. This, along with the interference from other users (which can be observed as a noise coloring eect), make impossible the use of conventional trellis-based turbo equalization.This problem is addressed in where compact receive antenna arrays and broadband beamformers (space-time lters) are employed for interference suppression and channel shortening so as to make possible subsequent use of trellis-based turbo equalization. The large computational load of conventional turbo equalizers, even with small channel lengths, makes their use unfeasible for MIMO wireless systems which require a separate turbo-type receiver for each transmit signal.

For this reason an alternative turbo space-time equalization scheme consisting of a broadband beamformer at its frontend and followed by a soft-input soft-output (SISO) decoder at its back-end. In the proposed receiver a soft interference cancellation operation is performed before beamforming using a priori symbol expectations and then the beamformer targeting a particular user employs minimum mean-square error (MMSE) equalization. Moreover each element of the beamformer decision sequence is mapped onto extrinsic symbol probabilities so as to be used by the SISO back-end decoder. Soft-interference cancellation, beamforming and probability mapping al together form a composite SISO module that is suitable for iterative processing. 

The design considerations and simulation results for turbo equalization of broadband signals with 8-PSK trellis coded modulation (TCM). Note that our turbo equalizer design is an extension of the linear ltering approach of to high-order signal constellations and diversity receivers within a MIMO transmission context. Because the proposed receiver does not rely on any trellis-search techniques for channel equalization, it oers signicant reduction in computational complexity over conventional iterative approaches using maximum a posteriori (MAP) equalization.