Hybrid Digital And Analog Processing In Massive MIMO

With the rapid development of mobile communication services,the current spectrum resources available can no longer meet the needs of future wireless services.Hence,making the most of millimeter wave(mmWave)band has become one of the main approaches to this problem.MmWave-band communication can increase the capacity.However,how to improve the reliability of mmWave-band communication becomes a challenge in designing practical systems.Combined with massive multiple-input multiple-output(MIMO)technology,the reliability of mmWave-band communication is greatly improved.The mmWave technology also solves the configuration difficulties of massive MIMO.In traditional massive MIMO,the use of fully digital precoding requires the configuration of large-scale RF chains,resulting in high power consumption and hardware costs.In order to reduce power consumption and hardware costs,there are currently two main methods.One approach is a new system architecture with a small number of RF chains;Another method is to use low-precision digital-to-analog converters(DACs)/ analog-to-digital converters(ADCs).When using a small number of RF chains,hybrid digital and analog processing is required.The performance may vary greatly compared to fully digital processing.Therefore,it is necessary to study the performance of massive MIMO systems under hybrid transceiver technology.This thesis focuses on massive MIMO systems,and studies the system performance under hybrid transceiver technology and wideband mmWave channel estimation under hybrid architecture and low-precision ADCs.Firstly,this thesis focuses on massive MIMO with limited RF chains and investigates hybrid digital and analog processing and the performance under certain hybrid methods in the second chapter.For massive multi-user MIMO downlink,the thesis gives the phase-aligned hybrid precoding scheme design and studies the system performance.The thesis then optimizes the design of hybrid precoding architecture,propose a novel phase-shifting and switching network based hybrid precoding structure,and provides a hybrid precoding design method under the architecture,which further reduces the power consumption of the hybrid precoding circuit.For the massive MIMO relaying with limited RF chains at the relay,the thesis further studies hybrid transceiver technology at the relay and the system performance under this technology.Secondly,This thesis considers the uplink of massive MIMO relaying with limited RF chains at the base station(BS)and studies hybrid digital and analog detection scheme at BS in the third chapter.In this scheme,the phase of each element of the hybrid analog detection matrix is selected as that of the corresponding element of the cascaded two-hop channel matrix.The thesis provides the performance in closed form under ideal phase shifters and quantized ones,respectively,which applies to arbitrary SNR,and then obtains the asymptotic performance under different SNRs at the relay and BS.The thesis then explores the potential of the system to reduce the transmit power of users and the relay.When the number of antennas increases,under the premise that the sum rate is monotonically increasing,the transmit power of users and the relay can be reduced by the exponential law of the number of antennas simultaneously.This thesis further investigates the multi-user optimal power allocation(PA)which maximizes the sum rate.The thesis converts it into an equivalent problem by introducing an auxiliary variable,and obtains the optimal PA coefficients in closed form by KKT analysis.Finally,this thesis considers wideband mmWave massive MIMO and focuses on channel estimation under hybrid-structure transceiver with low-precision ADCs in the forth chapter.By the linear modelling of nonlinear ADC quantization based on the Bussgang theorem,The problem is transformed into a statistically equivalent linear estimation problem which sequentially optimizes the frequency-flat analog weights and frequency-selective digital weights.Without any priori channel directional information,the thesis designs the analog weight matrices as isotropic Gaussian ones.After obtaining the statistics of the equivalent noise,the thesis derives the optimal digital estimating weight matrix in the sense of mean square error in closed form.The proposed channel estimator applies for arbitrary channel models.If a priori sparsity presents in mmWave channels,the proposed estimator further reduces the quantization noise caused by low-precision ADCs by additionally applying orthogonal matching pursuit.Simulation results verify that the proposed channel estimator outperforms the traditional ones under different ADC quantization bits.