Research On Beamforming Techniques In The High-speed Mobile Communication

Mutiple antenna beamforming is an important research area for the wirelesscommunications.By designing a precoder matrix with a partial or full knowledge ofchannel state information (CSI) at the transmitter, beamforming can obviously improvethe reliability and efficiency of multiple input multiple output (MIMO) or multiple inputsingle output (MISO) communication systems, without increasing the cost of extrabandwidth resource.A lot of researches on this topic have been made.Howerver, when itcomes to the high-speed mobile communication, the channel usually becomes frequencyand time selective.It will lead to several new problems:In frequency selective channel,orthogonal frequency-division multiplexing (OFDM) technology is often used totransform the frequency selective channel into several flat sub-channels.In this case,beamforming factors have two dimensionalities:space and frequency,which makes thedesign problem become more complexed; In time selective channel, it is difficult fortransmitter to acquire the exact CSI, which will have a severe influence on theperformance of transmit beamforming system. It’s crucial to overcome the influencebrought by the channel errors. The dissertation focuses on these problems, and the mainwork of this dissertation is listed as follows:Firstly, this dissertation addresses the beamforming design problem for cooperativedecode-and-forward (DF) system based on OFDM over frequency-selective channels.Adapting to the distributed relay network, an asymptotic optimal algorithm is proposedunder per-antenna power constraint (PPC). The simulation results show that theproposed algorithm significantly outperforms the existing PPC solutions and is robuston the different antenna power allocation and antenna number.Secondly, the beamforming design problem for cooperative amplify-and-Forward(AF) system based on OFDM over frequency-selective channels is discussed. In orderto get the minimum Bit-Error-Ratio (BER) at receiver, a two-stage method to designBF vector is proposed: At the first stage, the optimal frequency beamforming vectorsper sub-channel of OFDM based on maximization of Signal-to-Noise Ratio (SNR) aregiven, which makes the virtual beamforming be in the optimum direction. At the secondstage, we adopt the minimization of the simple arithmetic mean of the effective errorprobability (ARITH-EEP) as our criterion to allocate the power among sub-channels.The two-stage method can efficiently reduce the difficulty and complexity of thebeamforming vector design in the systems with many relays. Simulations show thatproposed algorithm significantly outperforms the existing methods in terms of BER.At last, the performance of MISO beamforming system with feedback delay andchannel estimation errors on time selective channels is analyzed.In the dissertation,Kalman-filter-based channel prediction is applied to follow the current CSI. Considering the analyses, a switch method based on critical SNR is proposed. Morever,the dissertation discusses the power allocation of channel prediction, and spatialcorrelation channel’s influence on the system performance.