Research On Hybrid Precoding Algorithm For Millimeter Wave Massive MIMO System

With the continuous improvement of people’s life level and the explosion of global application technology,the public has put forward higher requirements for high transmission rate and high reliability transmission.Millimeter wave communication technology is a new type of wireless communication technology,which can use high frequency millimeter wave to achieve large capacity and high reliability data transmission.This technology is foreseeable to be applied to future mobile communication and wireless communication fields such as Wireless Local Area Network(WLAN).Massive Multiple-Input Multiple-Output(Massive MIMO)is one of the core technologies of wireless communication,and hybrid precoding technology can adjust the direction of the signal to align the receiver direction,so as to improve the spectrum efficiency of the system and reduce the overall power consumption.Currently,there are two issues.On one hand,the system complexity is high due to the large number of antennas and the multi-variable joint optimization of hybrid precoding.This poses certain problems in the actual application of cellular communication.On the other hand,the trend of future networks is to have smaller coverage areas,thus more access points will exist in the unit communication area,resulting in serious interference problems.If conventional interference management methods are adopted in Massive MIMO and dense deployment AP schemes,the system complexity will be high due to the need for global channel state information.This thesis proposes the following two algorithms to address the above issues.In chapter 3,we propose a mixed precoding algorithm based on orthogonal matching tracking to extract the inverse matrix phase alternating minimization solution to solve the high computational complexity problem in point-to-point millimeter wave communication.This algorithm preprocesses by orthogonal matching algorithm,and obtains a local optimal solution under the minimum square criterion and the constraint of modulus.The simulation results show that when using the algorithm proposed in this thesis,the complexity is reduced by about sixtyseven percent compared with the traditional S VD decomposition extraction phase scheme,while ensuring the system spectrum efficiency and energy efficiency.In chapter 4,we propose a deep learning mixed precoding technology based on local channel information.Through the network characteristics of deep learning,it can effectively suppress the interference between highdensity APs and improve the achievable rate of the system.Through system-level simulations of multiple APs,this thesis proves that compared with the acquisition of global interference channel information,the proposed algorithm can maximize the system spectrum efficiency while greatly reducing the system-level coordination cost between APs and greatly reducing the system computational complexity.