Research And Performance Evaluation Of Massive MIMO Channel Equalization Technology

Massive MIMO technology refers to a transmission system that equips a large number of antennas at base station, services multiple users simultaneously, and can transmits data at a high rate and large-capacity. Massive MIMO technology which has become a hot topic at home and abroad is considered one of the key technologies of the future 5G communication. The thesis mainly focuses on pre-coding and channel equalization of Massive MIMO.First of all, the article introduces the latest research progress, the current major research focus and the technical difficulties. Besides, the thesis analyzes the system model and spectrum efficiency of Massive MIMO.Secondly, the thesis describes two common pre-coding algorithms: ZF(Zero Forcing) and MMSE(Minimum Mean Square Error) and comparatively analyzes their error performance and complexity. Besides, the chapter introduces pre-coding algorithm of MRT(Maximal-Ratio Transmit)-Schmidt which is based on the orthogonal transmit vectors. This algorithm selects all of the transmission vectors that need to be orthogonalized by a reasonable threshold. The pre-coding algorithm of Schmidt gains better pre-coding performance by getting the transmit vectors to be orthogonalized. MRT(Maximum- Ratio Transmit)-Schmitt pre-coding algorithm greatly reduces the computational complexity by getting the MRT pre-coding vectors to be orthogonalized, no matrix inversion.The chapter presents the concept of channel equalization and derivation. The thesis analyzes several common channel equalization algorithms: ML(Maximum Likelihood); ZF(Zero Forcing); MMSE(Minimum Mean Square Error); OSIC(Ordered Serial Interference Cancellation); LR(Lattice Reduction), compares their BER performance and complexity. Furthermore, the thesis describes the properties of channel hardening. Based on this feature, we can get the Inversion of the channel matrix by using the way of series expansion which has lower complexity. A new equalization algorithm: M-LAS(Multistage Likelihood Ascend Search) which has low complexity when the number of antenna is large has been introduced in the thesis. Based on the conventional M-LAS(Multistage Likelihood Ascend Search) equalization algorithm, the thesis introduces a new M-LAS equalization algorithm(Multiple Initial Vectors LAS, MIV-LAS)that has more than one initial value. MIV-LAS equalization algorithm is implemented by a local search in M set of values without matrix inversion, which greatly reduces the computational complexity. The proposed MIV-LAS equalization algorithm gets on multi- level LAS(Likelihood Ascend Search) search by using multiple initial values,which improves the BER performance.Finally, the paper realizes part of 64?8 MMSE equalization algorithm on the Xilinx K intex 7 chip.We analysis the error between FPGA implementation and the Matlab simulation. The error is within the acceptable range. The result is right on the Xilinx Kintex 7 chip and resource consumption and timing are within the normal range.