The Channel Estimation For Cell-Free Massive MIMO Based On LEO Satellite Mega-constellation

The development of a new generation of information technology prompted a diversification of mobile device trends and a higher level of digitisation,and people are demanding higher network speeds.The current MIMO(Multiple-Input Multiple-Output,MIMO)systems will not be able to meet the high demand for network speed in the near future.With the change of technology,Cell-Free architecture has been proposed.Cell-Free massive MIMO technology is an improvement of massive MIMO technology,whose main advantage is to eliminate cell boundaries and provide better service to users at the edge of the cell,which is one of the most promising technologies in the development of 6G.On the other hand,with the development of small satellite technology in recent years,it has become a reality to build a constellation of hundreds or thousands of LEO(Low Earth Orbit,LEO)ultra-dense satellites to provide a global coverage communication network.Based on this,this thesis proposes a Cell-Free massive MIMO network architecture for the LEO satellite Mega-constellation by integrating the LEO satellite Mega-constellation and Cell-Free massive MIMO technology.To keep in line with the Cell-Free concept,it focuses on the deployment of the LEO satellite Mega-constellation in the space segment and the channel estimation between the SAP(Satellite Access Point,SAP)and the ground terminal.In order to be consistent with the Cell-Free concept,the research on Cell-Free massive MIMO channel estimation for LEO satellite Mega-constellation is carried out around the two key issues of deployment of LEO satellite Mega-constellation in space segment and channel estimation between SAP and ground terminal,and the specific work is carried out as follows:(1)Analysis of the Cell-Free massive MIMO network architecture for the LEO satellite Mega-constellation.In keeping with the Cell-Free concept,the number of satellites covering an area is first determined,while the satellite coverage times are analysed to ensure that the SAP deployment is reasonable.Secondly,the channel characteristics in LEO communication scenarios are analysed and the channel model between the SAP and the UT(User Terminal,UT)on the ground is determined,and finally the main traditional channel estimation algorithms are presented.(2)A channel estimation study is carried out based on the previously modelled channel model.A convolutional neural network-based channel estimation algorithm is proposed to address the problems of low estimation accuracy and high complexity of traditional channel estimation algorithms.Specifically,the algorithm takes the initial value of LS(Least Squares,LS)estimation as the input data of the neural network,the first layer of the neural network is the channel matrix data value,the convolutional neurons are combined to extract the channel data features,the rectified linear units are used as the activation function between the layers of the network,then the convolutional layer finally passes the extracted data features to the fully connected layer,and finally the output channel estimate of the neural network.The aim of the neural network is to continuously update the weights and bias values of the convolutional neural network so that the output is as close as possible to the labelled value.Simulation results show that the proposed CNN(Convolutional Neural Network,CNN)channel estimation outperforms the traditional LS,MMSE(Minimum Mean Square Error,MMSE)and LMMSE(Linear Minimum Mean Square Error,LMMSE)algorithms.(3)The complexity of the aforementioned channel model makes channel estimation more difficult,combining mm Wave(Millimeter Wave,mm Wave)technology and Cell-Free massive MIMO technology,Cell-Free mm Wave massive MIMO is proposed to convert the complex.The proposed model converts a complex spatial channel into a mm Wave beam channel and then performs channel estimation using compressed sensing techniques.Specifically,using the sparsity of the mm Wave beam spatial channel,the channel estimation can be described as a sparse signal recovery problem,which can be solved by classical compressive sensing techniques.In order to improve the estimation accuracy,the SD(Support Detection,SD)channel estimation algorithm based on compressive sensing is proposed in this thesis.The simulation results show that SD algorithm outperforms the OMP(Orthogonal Matching Pursuit,OMP),SMD(Sparsity Mask Detection,SMD)and achieves higher estimation accuracy.