Research And Optimization On Low-complexity Detection Algorithms Of Massive MIMO Systems

With the booming development of Mobile Internet,the number of users and the amount of data transmission has been increasing sharply.The service provided by 3G/4G gradually cannot meet the demands of users and research on the 5th generation mobile communications(5G)has already begun.Massive MIMO technology which is the core technology of future 5G mobile network,can provide a remarkable increase in spectral efficiency and energy efficiency by using the large excess of station antennas.However,the increasing number of base station antennas lead to the increasing complexity of the detector,which becomes the bottleneck problem of massive MIMO application.This thesis mainly studies uplink detection algorithm of massive MIMO systems,including linear detection algorithms and low-complexity non-linear detection algorithms,especially an EP detection algorithm which is suitable for high-order massive MIMO systems.Firstly,this thesis uses the model of massive MIMO to analyze and compare linear detection lgorithms i.e.MRC,ZF and MMSE.The result shows that the simplest linear detection algorithm can achieve optimal performance when the number of receiving antennas is far larger than the number of transmitting antennas and grows infinite.However,the convergence to this behavior has been shown to be slow requiring that the number of transmitting antennas is significantly smaller than the number of receiving antennas.Therefore,it is strongly necessary to study low-complexity non-linear detection algorithm for massive MIMO systems.Secondly,this thesis simplifies the single-user MIMO model by assuming m(28)n,proposes a low-complexity high-accuracy EP algorithm for high constellation order,high dimensional MIMO systems.EP algorithm detects signal by constructing a gauss tree to approximate the posterior distribution p(7)u y(8)of transmitted symbol vector.This thesis has explored the principles and computational complexity of EP algorithm in detail and compared EP,MMSE-SIC and GTA-SIC detection algorithm.The simulation results show that the EP algorithm is robust and fast:Regardless of the number of base station antennas or the modulation order of the system,the EP algorithm can always be far superior to GTA-SIC and MMSE-SIC in performance,but only in several times computational complexity.