Sparse Channel Estimation For Reconfigurable Intelligent Surface Assisted Mm Wave Communication Systems

As one of the key technologies of the 6th Generation Mobile Networks(6G),Reconfigurable Intelligent Surface(RIS)has the advantages of lower hardware costs and energy consumption,and significantly improving coverage and capacity,which has been attracted high attention from academia community.In RIS assisted millimeter wave communication systems,obtaining accurate Channel State Information(CSI)through channel estimation is very important,as accurate CSI can better design the precoding matrix and RIS reflection coefficient,improving the quality of information transmission.However,composed of a large number of reflective components,RIS has no ability to process information and can only cause phase shift of the incident signal,resulting in a high CSI dimension of the system and an additional cost of pilots signal.To address this issue,the following research has been conducted in this thesis:Firstly,modeling the uplink information transmission process of RIS assisted millimeter wave single input multiple output(SIMO)communication system.Assuming that BS and RIS are fixed,and there is a partially identical path between RIS and different users.Analyzing the correlation channels among different users,a channel estimation method based on position information orientation is proposed.Greedy algorithm is first used to obtain a set of non-zero indices for different users,then,the least squares algorithm is used to estimate the specific parameters of the cascaded channel between different users.By combining greedy algorithm and least squares,the pilot cost is effectively reduced and accurate channel state information is obtained.Secondly,a channel estimation scheme based on a Controllable Arbitrary-pilots Deep Iterative Network is proposed.Construct a learnable deep iterative network and introduce a random prediction strategy to adapt to the diversity of pilot numbers.Users can send different numbers of pilots,and the deep iterative model can still estimate channel parameters based on the receiving signal,improving the generalization ability of network;Combining the traditional channel estimation method with the deep iterative network,the optimization process is transformed into the network iterative process.And using network learning instead of manually setting regularization parameters.The simulation verifies that the depth iterative network constructed can adapt to the changes in the number of different pilots and obtain accurate channel state information.Finally,the downlink information transmission of the RIS assisted millimeter wave multiple-input multiple-output(MIMO)communication system is modeled.Using tensor decomposition,a high-dimensional channel estimation problem is transformed into three independent two-dimensional channel estimation subproblems.Combined with the principle of compressed sensing,a Bilinear Alternating Iterative Shrinkage/Thresholding Algorithms based on tensor decomposition is proposed to estimate the channel between BS-RIS and RIS-UT.Then obtained the channel state information of cascaded channel,and verified the feasibility and effectiveness of the Bilinear Alternating Iterative Shrinkage/Thresholding Algorithms based on tensor decomposition through simulation.