Turbo space-time equalization of broadband wireless channels

We present space-time turbo (iterative) equalization methods for complex trellis coded modulation signals transmitted over broadband wireless channels, which is characterized by severe multipath fading. The multipath fading channel, protected by a TCM encoder/decoder pair, is described as a partially observed complex convolutional encoder and the received signal is viewed as the output of a serially concatenated coding system. It is well-known that in the narrowband transmission case, the equivalent encoder model for the channel can be represented by reasonable complexity and optimum symbol detection is achieved by iterative equalization/decoding schemes where soft-information is shared between the constituent decoders. However, for fixed wireless systems operating at high data rates, the multipath delay spread becomes large which makes direct application of trellis based equalization methods impossible.; The equalizer, proposed in the first part of this thesis, consists of a broadband beamformer which processes antenna array measurements to shorten the observed channel impulse function, followed by a conventional scalar turbo equalizer. Since the applicability of trellis based equalizers is limited to additive white noise channels, the beamformer is required to preserve the whiteness of the noise at its output which is equivalent to requiring that the FIR beamforming filters must have a power complementarity property. The power complementarity property imposes non-negative definite quadratic constraints on the beamforming filters, so the beamformer design is expressed as a constrained quadratic optimization problem. The composite channel impulse function at the beamformer output is shortened significantly, making it possible to use a turbo equalizer for the joint equalization and decoding of trellis modulated signals.; In most broadband systems, the transmission not only suffers from the intersymbol interference inherent to the multipath channel, but also from co-channel interference due to the presence of other users. The equalizer proposed in the second part interprets the presence of co-channel interference as a noise coloring effect and employs a broadband beamformer to jointly shorten the channel impulse response and whiten the channel noise. The noise whitening requirement again imposes nonnegative definite quadratic constraints to the beamformer design. The beamformer output sequence which takes the form of a shorter channel impulse function observed in the presence of additive white noise is applied to the turbo equalizer for joint equalization and decoding of modulated signal.