Research On Precoding Algorithms In Massive MIMO Systems

Massive multiple-input multiple-output(MIMO)which using large-scale antenna array at the base station to get more diversity gain can significantly improve the system capacity and energy efficiency,which is considered as one of the most key technologies to the fifth generation(5G)communications.However,the computational complexity,power consumption,etc.will increase sharply with the increase of the number of base station antennas.Therefore,a key requirement on the downlink in massive MIMO systems is to design low-complexity,low-energy consumption and high-performance precoding algorithms.There are three key research areas in this thesis: 1)studying in low-complexity and highfeasibility linear precoding algorithm;2)researching on low peak-to-average power ratio(PAPR)nonlinear to save cost and energy consumption,as well as effectively reduce the nonlinear distortion of signals;3)in view of the specialty of millimeter wave massive MIMO systems,studying in more applicable digital and analog hybrid precoding methods.The main research and achievements around the above three are listed below.1.Low-complexity linear precoding algorithms are introduced and studied.Firstly,a massive MIMO linear precoding system is presented.Then,a brief introduction about three classical linear precoding methods is given,includes maximal ratio transmission(MRT),zero forcing(ZF)and regularization zero forcing(RZF).And comparing their performances in detail by simulation.Finally,considering that RZF precoding has the better performance but involves unfavorable complicated matrix inversion,a low-complexity near-optimal RZF precoding scheme based on Gauss-Seidel method(RZF-GS)is proposed.After that,to further accelerate the convergence speed,a faster convergence initial solution is proposed.The proposed method can approach the capacity and BER performance of the RZF precoding with low computational complexity.2.Considering that linear precoding methods can lead to transmit signals with high peak-toaverage power ratio(PAPR).For this,it is necessary to use high cost and high consumption linear divices,such as power amplifier(PA),which can severely reduce the feasibility of the massive MIMO system.Therefore,nonlinear precoding schemes which yields low-PAPR transmit signals would be desirable to research,meanwhile,the system performance need to be ensured.In this aspect,a multiuser massive MIMO nonlinear precoding system is presented at first.Then two low PAPR classical precoding algorithms,constant envelope(CE)and annulus constrained(AC),are given,as well as their rapid convergent iterating solutions.Finally,to make up for the shortcomings of the above two methods,massive MIMO channel characteristics are taken into account and a low PAPR approximate message passing(LP-AMP)algorithm based on belif propagation(BP)is proposed.The proposed LP-AMP method can get PAPR reduction and enough transmission performance simultaneously with low computational complexity.3.Studying how to improve the system feasibility while massive MIMO and millimeter wave(Mm Wave)are used in conbination,which is focus on digital and analog hybrid precoding algorithms that are more suitable for Mm Wave massive MIMO systems.Therefore,first of all,a hybrid precoding system architecture based on fully connected is presented.Then,orthogonal matching pursuit(OMP)algorithm with near optical performance and low-complexity beam steering(BS)algorithm are studied in detail.Finally,an asymptotically unitary hybrid precoding(AUHP)algorithm is proposed,which takes into acount that the antenna array response matrices in base station are asymptotically unitary.Meanwhile,a more quickly convergent receive matrix solving scheme based on compressive sampling matching pursuit(CSMP)is given.