Research On Algorithm For Precoding In 5G Mm Wave Massive MIMO Systems

Recently,the communication technology has developed rapidly,which provides richer com-munication services and makes people's life more convenient.The requirements of communication services grow fast,especially the transmitting rate.There is no doubt that it will speed up the de-velopment of communication technology.But there are still some problems we need to solve,for example,the spectrum shortage problem.To solve this problem,in conventional orthogonal communication systems,higher order mod-ulation is used to improve spectrum efficiency.However,the high-order modulation systems is more sensitive to the transmitting noise and the non-linear response of the elements,which will lead to performance degradation.Recently,in non-orthogonal communication systems,people have resorted to the non-orthogonal techniques,e.g.,fast-than-Nyquist(FTN),whose complexity is much higher.As we can see,the aforementioned methods can not solve the problem of spec-trum scarcity directly.In order to solve this problem fundamentally,people try to figure out how to communicate in higher frequency,and the development of the communication techniques in new spectrum band have drawn great interest.Hence,lots of researchers focus on the transmitting techniques in millimeter wave(mmWave)band.Besides,due to its short wavelength,the size of antenna element can be small.Then mmWave communication techniques can work well with the massive multiple-input multiple-output(MIMO)systems.Aiming at.improving the system transmitting rate,this thesis focuses on the precoding opti-mization algorithm for the uplink of multi-user massive MIMO systems and the uplink of multi-user massive MIMO mmWave systems.Besides,the precoding for single-user non-linear mmWave communication systems is also considered.First,this thesis studies the precoding algorithm for the uplink of multi-user massive MIMO systems.In the system model presented in this thesis,several users communicate with a single base station(BS).Two kinds of messages,common messages and private messages,are considered,which are processed by different precoding matrices.In order to improve the energy efficiency,the BS employs antenna selection scheme,which means it only selects a few antennas to receive the messages.Hence,we aim at maximizing the system throughput by optimizing the precoding matrices of users,the antenna matrix and the receive matrix of BS.To address this problem,the Frank-Wolfe algorithm and the weighted minimum mean-square error(WMMSE)algorithm are jointly used.Besides,due to the sparsity of the antenna selection problem,we proposed a sparse algorithm based on quadratic approximation.The detailed proof of convergence is presented.As we can see from the simulation results,the proposed algorithm converges to a KKT point,and its performance approaches the exhaustive search method which is optimal.By solving the sparse optimization problem,a sparser antenna selection matrix is obtained,which leads to high system throughput.Second,we focus on the hybrid precoding algorithm for the uplink of multi-user massive MIMO mmWave systems.Since such systems are operated in mmWave band,we have to consider the hybrid precoding instead of the traditional fully digital precoding.In order to solve this hybrid precoding optimization problem,the concave-convex procedure(CCCP),the WMMSE algorithm and the Manifold optimization(MO)are jointly employed.By optimizing the digital precoding matrices of users,the antenna selection matrix,the analog receive matrix and the digital receive matrix,the system throughput is maximized.The convergence proof of the proposed algorithm is provided.From the simulation results,the two loops in the proposed algorithm both converge to KKT points,and the performance of the proposed algorithm approaches that of the method using fully digital receiver and exhaustive search method in terms of system throughput.Moreover,the proposed algorithm can achieve better energy efficiency.Last,the hybrid precoding algorithm is investigated for single-user massive MIMO non-linear mmWave systems.Considering the low energy efficiency of traditional linear elements in mmWave communication systems,a lot of non-linear elements are designed for communication systems operated at high frequency.Hence it makes sense to study the precoding algorithm for non-linear systems.In this thesis,the power consumption model for single-user massive MIMO non-linear mmWave systems are analysed,and the corresponding optimization problem is presented.To address this problem,the penalty dual decomposition(PDD)framework are exploited.The original problem are separated into several subproblems,which are easy to solved.The convergence of the proposed algorithm is analysed.From the simulation results,the PDD algorithm can converge to a KKT point and achieve good system throughput.It can be viewed as one of the potential design for the non-linear mmWave communication transceivers.