Research On Precoding And NOMA For Mmwave System

In order to satisfy the increasing demand of communication network system capacity,the wireless communication band has been very crowded,mmWave communication is a mature technology to improve the frequency utilization efficiency due to the small wavelength of mmWave.On the other hand,mmWave system has large number of antennas,and the cost of hardware is expensive,which offer challenges to mmWave system.In this thesis,we discuss three problem: hybrid precoding,initial access and the joint of power allocation and beamforming in mmWave NOMA system.In this thesis,we study a lot of literature.Besides,we improve the existing methods in aspect of throughput and computation.To sum up,major work and contribution are as follow:1.With regard to hybrid precoding,the spatially sparse precoding method which apply to full-array precoding structure adopt codebook,which need some assumptions on the antenna array geometry.We propose a no-dictionary two stage precoding method with sub-array precoding structure.The method has less hardware and computational complexity.When the power consumption of full-array structure and subarray structure is equal,the performance of the proposed method is higher than spatially sparse precoding method.2.With regard to initial access,we discuss the situation of multiply bases and multiply users.In this situation,multiply users may access in system at the same time.sequential stochastic approximation method has poor convergence performance.We propose a particle swarm initial access algorithm,which can escape from local minimum points by foraging act of research and observation about groups of birds.3.With regard to the joint of power allocation and beamforming in mmWave NOMA system,the thesis proposed by others considers a special mmWave NOMA systems where BS transmits single beam with two users.We propose a novel joint design of power allocation and beamforming for mmWave NOMA network with multiple beams.This problem is a non-convex problem and may not easily be convert to a convex problem.In order to make the problem tractable,a codebook is introduced in this problem.Further,we propose an iterative algorithm using the concept of successive convex approximation to solve the problem.In statistical sense,the algorithm complexity is less than that of codebook exhaustion method.