Research On The Low Complexity Signal Detection Technology In Massive MIMO Systems

Massive Multiple Input Multiple Output(MIMO)technology improves the spectral efficiency and the capacity of system by equipping hundreds or thousands of antennas.However,the sharply increasing number of antenna brings enormous pressure to massive MIMO system in terms of computational complexity,hardware cost and energy consumption,which hinders the commercialization of massive MIMO technology.Therefore,it has become a research focus to propose corresponding low-complexity signal detection techniques for different scenarios of massive MIMO application.The low-complexity signal detection technology is studied in this thesis,which respectively focuses on the spherical wave channel characteristic appearing in massive MIMO system and massive machine type communication(m MTC)system with low-resolution analog-to-digital converter.The main research contents are as follows:1.Considering the communication scenario that uniform circular array antennas are equipped at both ends of the transceiver,and more accurate spherical wave channel characteristic are introduced to establish channel model.Firstly,based on the special circulant structure of the established channel model,the optimal transmit and receive direction containing spherical wave characteristic can be obtained only by a fast fourier transform and matrix-vector multiplication operation,based on the optimal transmit and receive direction,the closed-form expression of the upper bound of the ergodic capacity for the system is obtained.Finally,based on the optimal tansmit and receive direction,combined with the principle of sphere decoding,the computational complexity is further reduced,and a sphere search aided with spherical wave characteristic(SSASWC)detection scheme is proposed.The simulation results show that the proposed SSA-SWC scheme can achieve better performance than traditional detection scheme with a lower computational complexity.2.The uplink of a grant-free massive machine-type communication system faces more practical challenges,including low resolution quantization,correlated fading channel,and unknown activity rate of machine type device(MTD).Firstly,the generalized expectation consistent signal recovery(GEC-SR)algorithm is introduced.Then,combining Woodbury formula and Neumann-series to approximately calculate the inversion operation of the high-dimensional matrix in the GEC-SR,thereby reducing the computational complexity.Finally,taking advantage of the structured sparsity of frame,a approximation generalized expectation consistent based on multiple measurement vector(AGEC-MMV)algorithm is proposed.The numerical results verify that the proposed AGEC-MMV algorithm can not only make full use of the frame structured sparsity and discrete prior information to improve detection performance,but also can approximate the performance of the GEC-MMV algorithm with lower computational complexity.In addition,the proposed AGEC-MMV algorithm outperforms existing compressed sensing algorithm in terms of robustness.In summary,the results obtained in this thesis on the research of low complexity signal detection technology in massive MIMO system are of great significance.