On Precoding For MIMO Channels

In multiple input multiple output(MIMO)systems,the precoding schemes at the transmitter,which uses the channel state information(CSI),are proposed to improve the spectrum efficiency and the power efficiency.However,in practice,the CSI at the transmitter is obtained from the receiver,which results in imperfect CSI at the transmitter.This dissertation deals with the precoding problems in MIMO systems with imperfect CSI at the transmitter and the main contributions achieved are as follows.In MIMO systems,firstly,using the channel mean feedback model,we improve the traditional water filling algorithm and present a joint power allocation and adaptive modulation scheme.In this scheme,the power determined by the water filling algorithm is reallocated with the combination of adaptive modulaiotn.The discontinuous power allocation scheme determines the allocated power and modulation constellation on each subchannel simultaneously.Then,using the limied feedback model,we discuss the beamforming and power allocation for MIMO systems based on Grassmannian space packings.The codebook design of beamforming matrices is proved to be a problem of Grassmannian space packings and the beamforming and power allocation scheme is presented based on the MMSE criterion.Finally,since the Grassmannian space packings can only be applied on Gaussian channels,as for other channels, we present a codebook design method by using ameliorated Lloyd vector quantization algorithm.Additionally,the vector quantization algorithm is proved equivalent to Grassmannian space packings at high SNR.In MIMO-OFDM systems,firstly,the precoding for space diversity MIMO-OFDM systems is studied.A precoding scheme including beamforming,power allocation and adaptive modulaiton is presented.By using the beamforming codebook constructed according to Grassmannian space packing,the optimal beamforming vector is determined at the receiver and its index is conveyed to the transmitter.Then the power and bits are allocated to each subcarrier using 'greedy algorithm' at the transmitter. The feedback data are greatly decreased compared with the previous schemes while maintaining a comparatively large capacity.Then,we study the precoding for space multiplexing MIMO-OFDM systems.Here,we propose an interpolation-based vector quantization precoding scheme.In this scheme,the subcarriers are divided into clusters whose precoding matrices are selected in a codebook aiming at capacity maximization at the receiver.And the index of precoding matrix for each cluster is conveyed back to the transmitter,which obtains all the subcarriers' precoding matrices through interpolation.In this scheme,the codebook is designed through vector quantization medthod.At last,the precoding schemes for space diversity MIMO-OFDM systems is extended to multiuser space diversity MIMO-OFDM systems and a multiuser adaptive precoding scheme is provided.In space-time block coding systems,using the channel mean feedback model, the design of optimal beanforming filters under the minimum average bit error rate (BER)criterion is presented,and the power allocation method on each beam is discussed. When the feedback bandwidth is limited,the precoding scheme for STBCMIMO systems is proposed.The codebook design is proved to be a Grassmannian space packings problem and the precoding matrices selection strategy for the purpose of minimizing the upper bound of the symbol error rate(SER)is presented.Simulation results show the limited feedback precoding scheme outperforms the antenna selection scheme with the same number of feedback bits.Finally,the multiple rate modulaiton scheme based on ARQ protocol is studied.The throughputs of constant rate modulaiton and multiple rate modulaiton are deduced.Further,we prove the upper bound of the throughput of multiple rate modulaiton is Shannon capacity.