Research On Multidimensional Matrix Receiving Algorithm For IRS Assisted Cooperative Communication

With the rise of research in sixth generation(6G)mobile communication technology,the increase of hardware cost and the influence of obstacles are important problems to be solved urgently in the communication system.Regarding this problem,the recently emerging key candidate technology for 6G-intelligent reflecting surface(IRS)has become a potential efficient solution.In order to take full advantage of the IRS-assisted cooperative communication systems,it is essential to design accurate receiving algorithms,with the core focus on channel estimation and signal detection.However,the dynamic nature of the mobile user end results in a relatively fast timevarying rate of the channel,which limits the applicability of traditional quasi-static channel estimation methods in mobile scenarios.Moreover,in harsh environments,the prolonged use of IRS can cause hardware damage,such as array blocking,resulting in severe performance degradation in the receiving algorithm.In addition,the traditional methods for IRS-assisted millimeter wave communications only estimate cascaded channels,which reduces the accuracy of the receiving algorithm.Based on the above challenges,this thesis focuses on the development of an efficient receiving algorithm for an IRS-assisted communication system based on a multidimensional matrix model.The main research of this thesis is as follows:(1)Firstly,for the IRS-assisted time-varying MIMO cooperative communication system,a time-varying nested Kronecker product parallel factor(PARAFAC)rearrangement alternating least squares(TNKP-RALS)receiving algorithm is proposed.According to the time-varying characteristics,the corresponding time-domain protocol is proposed.The algorithm can construct the first hop time-varying channel into the form of tensor and estimate the mod-3 form of the first hop time-varying channel by the alternate least squares method which increases the fitting speed.Using the inverse conversion of the first hop channel mode-3,the time-varying tensor channel is recovered.The Kronecker rearrangement method is used to estimate the second hop channel and signal matrix.The algorithm can improve the accuracy of estimation and overcome the disadvantages of cumulative error in traditional algorithm.It reduces the number of convergence iterations.Secondly,to solve the IRS blocking problem,the array blocking error of IRS needs to be estimated.Therefore,a time-varying nested Khatri-Rao product and Kronecker product PARAFAC rearrangement factorization alternating least squares(TNKRKP-RFALS)receiving algorithm is proposed.The first hop time-varying channel estimation method of this algorithm is the same as the TNKPRALS algorithm.Then Kronecker rearrangement and Khatri Rao decomposition method are used to estimate the second hop channel,signal and IRS array blocking error matrix respectively.This algorithm can still guarantee the receiving algorithm performance without replacing the blocking IRS.Simulation results show that the normalized mean square error of the proposed algorithm is better than other receiving algorithms,and its bit error rate is close to that of the zero-forcing algorithm as the reference boundary.(2)According to the IRS-assisted multi-user millimeter-wave MIMO system,a two-stage PARAFAC-3 combined constrained PARAFAC-4 one-dimensional search and rearrangement alternating least squares(TP3CP4-ORA)receiver algorithm is proposed based on tensor decomposition.The external channel and internal parameters are estimated in two stages,which do not affect each other.In the first stage,the receiving signal is constructed as a three-order PARAFAC model,and the separated two-hop channel estimation values are obtained using bilinear alternating least squares algorithm.This stage overcomes the drawbacks of signal distortion and error accumulation in millimeter-wave cascaded channel estimation algorithms,and improves the flexibility of channel estimation.In the second stage,the received signal was constructed as a four-order structured-constrained PARAFAC model.The angle and path gain parameters were obtained by first using the quadrilinear alternating least squares algorithm and then using a one-dimensional search method and Kronecker rearrangement.The retrieval of internal parameters provided more information to optimize signal transmission,thereby improving system performance.Simulation results show that the mean square error of the TP3CP4-ORA algorithm can approximate Cramer-Rao Bound,and the performance of mean square error is better than the existing algorithm.