User Scheduling And Localization For Massive MIMO Systems

With the rapid development of information technology,information services have experienced the transformation from fixed access points to mobile terminals,and the transformation from providing text and voice services to providing multimedia and Internet access services for mobile users.With the increasingly tight wireless spectrum resources,how to further explore space wireless resources on the basis of the 4th generation(4G)mobile communication system and improve the spectrum efficiency and power efficiency of mobile communication systems has become an urgent problem to be solved in the upcoming 5th generation(5G)mobile communication system.Massive multiple-input multiple-output(MIMO)has become one of the promising technologies for the upcoming 5G wireless communications for its enormous potential in spectral efficiency and power efficiency.In frequency-division duplex(FDD)massive MIMO,the downlink training and feedback overhead scale linearly with the number of BS antennas and will overwhelm the precious downlink resources.Meanwhile,wireless localization for mobile devices is drawing increasing interest from both industry and academia as locationbased services(LBSs)spring up in recent years.This dissertation studies the agglomerative user clustering and cluster scheduling for FDD massive MIMO systems and the fingerprint based single-site localization method for massive MIMO frequency-division multiplexing(OFDM)systems.In order to reduce the overhead for the channel state information at the transmitter side(CSIT)acquisition and suppress the inter-cluster interference,an agglomerative user clustering and cluster scheduling scheme for two-stage precoding is proposed.The average signal-toleakage-plus-noise ratio(SLNR)based iterative cluster scheduling scheme combines the outer precoder design into the iterative cluster scheduling process to suppress inter-cluster interference while achieving the appropriate set of user clusters.For low complexity wireless localization,a fingerprint based single-site localization method is proposed for MIMO-OFDM systems by taking full advantage of the high resolution in the angle and delay domains.In order to achieve high localization accuracy as well as reducing storage overhead and computational complexity,a deep convolutional neural network(DCNN)enabled localization method is proposed,in which the modeling error for fingerprint similarity calculation can be overcome.A hierarchical DCNN architecture is also proposed for practical implementation.The major results and contributions of this dissertation are listed as follows.1.An agglomerative user clustering and cluster scheduling scheme for two-stage precoding is proposed.Two-stage precoding is considered in a more realistic scenario with active users randomly distributed in a cell.A new agglomerative clustering method to significantly simplify the user clustering process is proposed.The method consists of two stages,the agglomerative user clustering and the following average SLNR based cluster scheduling.By exploiting the bottom-up clustering process,the agglomerative user clustering method can be distinctly computational efficient and performance guaranteed without iteration.The average SLNR based iterative cluster scheduling scheme combines the outer precoder design into the iterative cluster scheduling process to suppress inter-cluster interference while achieving the appropriate set of user clusters.The lower bound of the average SLNR is extended to a more realistic scenario where users in the same cluster have different channel statistics.Furthermore,for the special case of uniform linear arrays(ULAs),a discrete Fourier transformation(DFT)based channel eigenspace approximation method is considered to significantly decrease the computational complexity in both user clustering and cluster scheduling.Simulation results validate the performance improvement of the proposed methods over the existing methods.2.A fingerprint based single-site localization method is proposed.Localization for mobile devices is drawing increasing interest from both industry and academia as LBSs spring up in recent years.For rich scattering environments,such as urban areas and indoor corridors,fingerprint-based localization techniques are very promising.For low complexity wireless localization,a fingerprint based single-site localization method is proposed for MIMO-OFDM systems by taking full advantage of the high resolution in the angle and delay domains.A new angle delay channel power matrix(ADCPM)fingerprint is extracted from instantaneous channel state information(CSI)by taking full advantage of the high resolution in the angle and delay domains for massive MIMO-OFDM systems.A new fingerprint similarity criterion is proposed to facilitate localization.Based on the new criterion,an efficient location estimation method is developed.To reduce storage overhead and matching complexity,a fingerprint compression method and a two-stage fingerprint clustering algorithm are also proposed for database preprocessing.Numerical results demonstrate the desirable performance of the proposed localization method.3.A fingerprint-based location for massive MIMO-OFDM systems with DCNNs is proposed.Fingerprint technique is a promising enabler for mobile terminals(MTs)localization in rich scattering environments,such as urban areas and indoor corridors.A fingerprint-based location for massive MIMO-OFDM systems with DCNNs is proposed.By taking full advantage of the high resolution in the angle domain and the delay domain in massive MIMO-OFDM systems,an efficient angle-delay channel amplitude matrix(ADCAM)fingerprint extraction method is first proposed.Then a DCNN enabled localization method is proposed,in which the modeling error for fingerprint similarity calculation can be overcome.Both DCNN classification and DCNN regression are considered.For practical implementation,a hierarchical DCNN architecture is proposed.The performance of the proposed DCNN localization method is evaluated via simulation performed in a geometry-based ray tracing signal propagation scene.Numerical results demonstrate that DCNN performs well in achieving high localization accuracy as well as reducing storage overhead and computational complexity.