Research On Techniques Of Channel State Information Acquisition And Precoding In Massive MIMO

Multiple input multiple output(MIMO) technology is becoming mature, and incorporated into emerging wireless broadband standards like LTE, and it serves mobile users with stable rates and ideal spectrum efficiency. In order to further develop the potential advantage of MIMO technology in the future wireless communication protocol,people has begun to focus on Massive MIMO technology. Massive MIMO systems can greatly improve the spectral efficiency and channel capacity, reduce the transmission power of base station and improve the quality of communication based on the spatial diversity gains brought by more antennas. To obtain the various benefits of Massive MIMO systems, however, the base station should have accurate channel state information(CSI). Many previous studies focused on the time division duplexing systems due to the convenience of channel reciprocity, but there are many frequency division duplexing(FDD) systems deployed worldwide and how can the base station obtain accurate CSI while reducing the overhead of channel estimation in Massive MIMO systems remains to be a problem. In addition, general linear precoding methods can achieve good gain in Massive MIMO, but reducing the precoding complexity and optimization is an urgent problem to be solved.In the first chapter, the paper summarizes the technical advantages of Massive MIMO compared to traditional MIMO, introduces the research status of channel state information acquisition and precoding, and points out the research direction of the paper.The second chapter introduces the channel characteristics of Massive MIMO, builds the mathematics model and discusses the channel spatial correlation and sparsity with explanations for the reasons to build the model. According to the channel correlations in time and space, multiple training is proposed in the third chapter. This training scheme breaks through the limitation of performance of the traditional single-shot training, and reduce the time overhead of pilot symbols.Meanwhile we propse two close-loop training schemes, which have better performace than open-loop training. Finally, based on multiple training, we study on the new pilot beam design method which has better performance compared with several pilot beam design methods proved by simulations.According to the joint sparsity of the multiuser Massive MIMO channel model, the fourth chapter proposes a joint CSI recovery algorithm based on compressed sensing(CS) at base station. The compressed measurements are taken locally at users, while CSI recovery is jointly performed at the base station. The proposed scheme has been shown to outperform other algorithms in both theoretical derivations and simulation results. In the fifth chapter, we improve the existing precoding method by using the matrix polynomial expansion to approximate the inverse of matrix. The complexity of the new precoding algorithm will not increase with the system dimension, and it fits for hardware implementation of parallel and pipeline, which makes the algorithm very attractive in Massive MIMO. The final sixth chapter makes the summary to the full text content, and point out the directions for further study and improve.