Abstract
In a wireless system with multipath MIMO (multiple input multiple output) channels using antenna arrays, the delay spread of multipaths results in intersymbol interference (ISI) and channel frequency selectivity. Similar to the channel frequency selectivity, spatial gains at the multipath angles also naturally result in channel angle selectivity. Based on the channel selectivity structures characterized by the path delays and the path directions-of-departure/arrival (DODs/DOAs), in this thesis, we first seek new insights into the matching of space-time codes and multipath MIMO channels in their angle-frequency (AF) structures. Next, by exploiting the wireless MIMO channel structure, new space-time code design criteria are derived. New structure-based space-time codes are identified through computer searches to justify the new criteria. Simulation results show that these codes have superior performance over the existing codes in the corresponding frequency-selective channels. Based on the new design criteria, we propose two low-complexity channel-adapted space-time (CAST) coding schemes, where trade-offs among codeword error rate, data throughput and computational complexity are very flexible. Simulation results confirm that, in the frequency-selective MIMO channels, the CAST coding schemes can perform significantly better than the existing space-time codes, e.g., Alamouti space-time orthogonal code. In addition, recent advances in information theory show that employing multiple antennas at both sides of a wireless link promises enormous capacity potential. With knowledge of channel state information (CSI) at the transmitter, space-time eigen-beamforming is the optimum coding scheme to exploit this potential. However, in non-stationary wireless environments, high complexity on MIMO channel tracking and large amounts of CSI feedback render such an approach impractical. By exploiting the wireless multipath channel structure, a space-time coding scheme involving a novel structure-based water-filling algorithm is proposed. Outage capacities evaluated through Monte Carlo simulations confirm the performance advantage of the proposed space-time coding scheme. For more efficient usage of bandwidth, blind detection schemes attract more and more attention. In a noncoherent system, a differential phase coding scheme is an attractive technique as it obviates the need for phase synchronization. However, its performance degrades considerably when channels vary rapidly. To establish a reliable communication link, we propose a wireless system with a blind receiver which jointly performs noncoherent channel estimation and serially-concatenated turbo code decoding over fast time-varying Rayleigh-fading channels. The low complexity blind receiver consists of two parts: 1) a Kalman filter as its channel estimation part and 2) two decoders, including a differential decoder and a convolutional decoder, as its signal decoding part. With various soft information, calculated in the maximum likelihood sense, iteratively passed around between the channel estimator and the signal decoder, the system is expected to hopefully approach the optimal performance. Note that, with no training data in the proposed system, it is impossible for the Kalman filter to avoid the CSI phase ambiguity problem, which can be perfectly taken care of by the differential decoder. Computer simulations confirm that the proposed system exhibits robustness against fast time-variation of Rayleigh-fading channels.