Research On Multiple Antenna In Milimeter Wave MIMO Systems

Millimeter wave (MMW) communication has attracted much attention as a key technology for next-generation WLAN and 5G cellular system. Compared with conventional low-frequency system, millimeter-wave communication suffers from severe propagation loss. So it has to employ large antenna arrays at both the transmitter and receiver, which can provide high beamforming gain and realize sufficient link budget. Beamforming technology is widely used in multiple antenna for improving signal to noise ratio, reducing error probabilities and increasing the system capacity.In beamforming, the key issue is to choose the optimal beam vectors. The thesis mainly discusses two beamforming technology for millimeter-wave orthogonal frequency division multiplexing (OFDM) systems and multiple-input multiple-output (MIMO)systems.Firstly, we proposed a codebook-based beamforming technology for 60 GHz millimeter-wave multi-antenna OFDM systems. Given a pre-specified codebook, the purpose of the beamforming technology is to find out the best transmit and receive beam vectors to maximize the spectral efficiency. We have to transform the above problem into the estimation of several sparse signal vectors by properly modeling the multipath structure in 60 GHz channels. Moreover, compressive sensing (CS)technology can estimate such sparse signals efficiently, wherein the measurement matrix can be designed by minimizing the inner products between different columns of the CS effective dictionary. As a result, CS-based iterative beamforming schemes are proposed in this paper. Due to the ability to sufficiently exploit multipath sparsity operating in 60 GHz channels, the proposed schemes significantly outperform the existing counterparts in terms of training overhead and success probability, which is affirmed by theoretical analysis and extensive simulations.However, only a limited number of radio frequency (RF) chains are available for MMW antenna arrays because of the power and cost limitation. In other words, the number of antenna elements is far larger than that of RF chains. Due to the requirement of realizing the optimal singular value decomposition (SVD) beamforming conventionally, this MMW MIMO setup poses a formidable challenge for channel estimation. To circumvent the formidable channel estimation, this paper proposes two iterative training schemes for the SVD beamforming in MMW MIMO systems.Depending on the channel reciprocity in time division duplex (TDD), the proposed schemes employ power iteration and Lanczos iteration, respectively, which can gradually approach the SVD beamforming. Besides, we only need to estimate several channel-related vectors instead of the MIMO channel matrix during iterations. Thanks to the spatial sparsity in MMW MIMO channels, the training overheads required by the proposed schemes are moderate. Simulations demonstrate that the proposed schemes outperforms the counterpart in terms of system capacity and training overhead, and can achieve the performance very close to that of the perfect SVD beamforming.