Research On Channel Information Acquisition And Beamforming For Millimeter Wave MIMO System

As the core general technology of the new round of scientific and technological revolution,5G has become the key technology to support the development of the economic society in the direction of digitalization,networking and intelligence under the background of the booming development of the Internet of things,the Internet of vehicles and artificial intelligence.Millimeter wave plays an indispensable role in 5G communication because of its abundant and unexplored spectrum resources which can be used to meet the requirements of greater bandwidth and ultra-high data rate.In order to compensate the strong loss of millimeter wave transmission in free space path,researchers propose to integrate large-scale MIMO into millimeter wave communication system to obtain higher beamforming gain and ensure the required communication quality.Compared with all digital or analog beamforming,a hybrid beamforming architecture,which includes analog and digital processing units has become a research hotspot in order to reduce the energy consumption and hardware cost of the system.However,it is harder to acquire channel state information with the decrease of RF chains and signal dimension,which will affect the performance improvement of beamforming.Thus,the joint channel estimation and optimal beamforming design in narrowband millimeter wave hybrid beamforming systems with massive MIMO are the focus of this paper.Since millimeter wave channels are sparse in time and beam domains,channel information can be obtained using sparse signal recovery techniques.A joint channel estimation and beamforming design scheme based on Manifold Alternating Direction Method of Multipliers is proposed in this paper.The large joint optimization problem is decomposed into several local sub-problems,and the multi-variable optimization is solved by successive iterative updating of each variable.In this scheme,optimal search methods based on Riemannian manifold and Oblique manifold are used to solve the discrete constant modulus constraint and the norm constraint during the beamforming design.Finally,the orthogonal matching pursuit algorithm is used to recover the channel information when the beamforming matrices converge to the optimal solutions.The results of simulation show the good performance of the scheme.A channel estimation and beamforming design scheme based on CB-ROMP is also proposed in this paper.The ROMP algorithm which can increase the resolution in angular search spaces of dictionary matrix is used to obtain the channel information based on the determined beamforming codebook.Due to the mutual coherence of sensing matrix is an important factor affecting the recovery performance of CS method,a general beamforming codebook design scheme based on minimizing the total coherence of the equivalent sensing matrix is proposed in this paper to ensure the accuracy of channel information acquisition.The simulation results show that the scheme has good and stable performance in several different analog beamforming structures.In addition,the relationship between low overhead and performance loss is explored and a beamforming codebook with optimal column permutation scheme suitable for low pilot cost to reduce the training overhead is proposed in this paper.The simulation shows that this scheme can achieve a lower mutual coherence and a trade-off between reducing training cost and performance loss.