| Multi-antenna system, named as Multiple-Input Multiple-Output (MIMO) system, which could improve transmission capacity and link reliability, has become one of key technologies that realize the high-speed broadband wireless Internet access networks in the future and has wide application prospect in the next generation mobile communications. This thesis discusses single-user downlink precoding algorithms based on spatial multiplexing; transmit diversity and hybrid scheme successively. In variance of CSI cases, the precoding theories are deduced, the algorithms are analysised, and the precoding performances are simulated respectively. The main contributions of the thesis are listed as follows:Firstly, a new precoding technique is proposed for STBC-VBLAST hybrid scheme. In principle of minimizing pair-error probability or maximizing channel capacity, the transmitter firstly decomposes the sub-channels for each STBC layer using QR-type decomposition; the non-interference equivalent sub-channel is the block matrix which is in the diagonal line of the upper-triangle R matrix. The eigen-beamforming direction of each layer is the eigen-vector of corresponding sub-channel. Then, the energy is loaded on every eigen-beaformer submitting to water-filling principle. All the sub-precoder are block-diagonalized to form system precoding matrix. This proposal is suitable for TDD mode. It takes advantage of CSI to reduce error rate and increase capacity. Matlab simulation results show that this algorithm could achieve remarkable effects espscially in low constellation dimension.Secondly, for the case when channel correlation and rank deficiency exist in spatial multiplexing VBLAST system and the channel covariance information is known by the transceiver. The pair-wise error probability in correlated channel is considered to design statistical precoder. It is concluded that the optimum statistical precoder is constructed by the domain eigenvectors of transmit correlation matrix, meanwhile, the total power is allocated to eigen-direction averagely. From the simulation results, it is observed that statistical precoder can degrade the performance attenuation caused by channel correlation.Thirdly, for spatial correlated open-loop linear constellation precoding (LCP) system, by analysis of the error rate performance and coding criterion, it is concluded that LCP scheme can provide diversity gain with higher coding gain in i.i.d. (independently and identically distributed) or lower correlated channels, but in higher correlated channels, because of strong correlation effect, the LCP scheme can only obtain diversity gain without any coding gain.This thesis also overviews precoding algorithms in different CSIT for the spatial multiplexing VBLAST system and transmit diversity STBC system.For VBLAST system, precoding in four interesting CSIT are discussed: 1) optimal linear precoding when the transmitter knows perfect CSI, the receiver employs linear or non-linear detector; 2) linear statistical precoding when the transmitter only has the knowledge of channel covariance in rank deficiency channel; 3) in limited feedback system, the transmitter chooses appropriate codeword from codebook as precoding matrix. Codebook designed by Grassmannian subspace packing and vector quantization are mainly introduced; 4) when the transmitter acqires channel covariance information and infinite rate feedback, the codebook is constructed by the original eigen-vectors of transmit correlation, rotated and scaled eigen-vectors. This performance is between statistical precoder and optimal precoder in perfect CSIT.For full transmit diversity STBC system, the optimal objective functions in different CSIT are derived from error performance separately. These CSIT cases consist of channel covariance information, mean and covariance information, and channel estimate error. Then, linear precoder is achieved by solving corresponding function. And the error performances in the aforementioned CSIT cases are simulated separately. |
PDF Full Text Request