Research On Combing And Precoding Algorithms In Cell-Free Massive MIMO System

As one of the key technologies of 5G,the Massive MIMO system has been greatly developed in recent years.Developing from the distributed Massive MIMO system,the Cell-Free Massive MIMO system gradually becomes one of the most promising structures for the next generation mobile communication system.The system comprises a very large number of distributed access points,which simultaneously serve a much smaller number of users within the over the same time/frequency resources.so as to be able to combat shadow fading and effectively suppress interference.However,this distributed architecture makes the channels between access points and users no longer satisfy independent and identical distribution,and many uplink and downlink signal processing methods in traditional centralized Massive MIMO systems are no longer applicable,so uplink and downlink signal processing is one of the issues that need to be studied in the current Cell-Free Massive MIMO system.Firstly,we describe the model of the Cell-Free Massive MIMO system,including its channel model and transmission protocol.The state-of-the-art uplink combining algorithms and downlink precoding algorithms are discussed,and then the uplink and downlink achievable spectral efficiency are derived.We analyze the deficiencies and problems within these uplink combining algorithms and downlink precoding algorithms,which inspire the following research.In comparision,we analyze spectral efficiency of the traditional Cellular Massive MIMO system.Then,the Cell-Free Massive MIMO system model with limited fronthaul capacity is then introduced,and the effect of the fronthaul capacity on the system uplink and downlink spectral efficiency is elaborated in detail.Secondly,to respond to the challenges mentioned above,we consider taking advantage of the deep convolutional network to improve the uplink combing algorithm and downlink precoding algorithm.We first introduces the theory of deep convolutional neural network and the idea of objective-oriented unsupervised learning.On this basis,we devise a new uplink combing and downlink precoding scheme.The spectral efficiency performance and the computational complexity of the proposed scheme and the traditional alogrithms are compared.Finally,the scalability of the traditional Cell-Free Massive MIMO system is discussed,and a user-centric system architecture is proposed to meet the system scalability conditions.Then we derive the uplink and downlink spectrum efficiency performance in this new system architecture.According to the characteristics of this usercentric architecture,an improved uplink combining and downlink precoding algorithm is designed,and the impact of the fronthaul capacity on its spectral efficiency performance is analyzed.Finally,we summarize this paper,and provide some unresolved problems and research directions in the uplink and downlink signal processing problems of the CellFree Massive MIMO system.