| To effectively improve the throughput of wireless systems, MIMO techniques have become attractive to many researchers. For MIMO point-to-point communications, sphere decoding provides optimal solution, however, it suffers from the variable high complexity incurred by the high dimensionality of transmit antennas and the difficulty to choose initial radius. This thesis develops an alternative approach, ordered tree traversal, to avoid the use of initial radius, and the ordered tree traversal algorithm damatically reduces the number of leaf nodes and the entire nodes explored during the candidate search process, since it only investigates the necessary leaf nodes. The possible drawback of the proposed method could be the memory cost. To further limit the computational efforts and storage space, the approximate methods, such as integer vector perturbation and differential SD/OTT are supplied as more options.;Still, in some scenarios, particularly where the ML solution becomes infeasible in practice due to the increase of the number of transmit antennas, then suboptimal methods based on an approximate solution to the lattice closest vector problem, called lattice-reduction-aided detection techniques, can be a good alternative. LLL lattice reduction algorithm is popular for its polynomial time solution. It alternately performs size-reduction and exchange steps until the algorithm converges. We introduce a new lattice reduction algorithm for MIMO systems. Basically, the Seysen's algorithm considers all the vectors simultaneously, instead of only the two adjacent ones in LLL. Seysen's algorithm can achieve BER performance improvement over LLL's with less computational complexity, which is measured by the average number of column operations required for lattice reduction before converge.;When perfect channel state information is available at transmitter, the so-called precoding methods were discussed. We unified three MIMO broadcast precoding techniques: Tomlinson-Harashima Precoding, Dirty Paper Coding and Vector Perturbation, by representing them in the same mathematical form. And the applications of sphere decoding and lattice reduction techniques are also illustrated at the transmitter side.;In the second part of this thesis, we studied the coded cooperative communications for downlink and uplink systems of 3GPP Long Term Evolution (LTE). The new cooperative strategies were proposed to exploit time diversity, space diversity (due to multiuser cooperation), frequency diversity and coding gain. Analysis on the pairwise error probability and the frame error probability shows that the new schemes have significant advantages over direct transmission, where both block fading channels with various block correlations, and the special case: quasi-static fading channel are under consideration. The numerical results are consistent with our theoretical analysis. |
