Research On Power Control And Decoding In Cell-free Massive MIMO System

As a new implementation of network MIMO,cell-free massive MIMO system has attracted wide attention since it was first proposed.Cell-free massive MIMO system comprises a large number of distributed access points over a large area,which will simultaneously provide services to all users.And they are connected to the CPU through the backhaul network.Comparing with conventional cellular network,cell-free massive MIMO performs better in macro diversity and multi-user interference suppression,for the inherits of channel hardening from massive MIMO and closer distance between users and access points.In addition,high energy efficiency and spectral efficiency can be enjoyed by cell-free massive MIMO system with power control strategy and user allocation scheme.Moreover,access points in cell-free massive MIMO can be accommodated in space-constrained environment.Therefore,comparing with centralized massive MIMO system,the deployment cost of cell-free massive MIMO system is lower.However,higher backhaul requirements are needed in cell-free massive MIMO to transmit data between access points and CPU for its distributed system framework,while finite backhaul resources will limit the performance promotion.By introducing the co-processing concept in a user-centric fashion,the demand of backhaul resources in cell-free massive MIMO has dropped.In addition,the actual coverage of access points in cell-free massive MIMO system is not sufficient to provide service for all users at the same time.Therefore it is of great value to do research on user-centric cell-free massive MIMO system.In cell-free massive MIMO system,the quality of channel estimation is one of the decisive factors that affects communication quality for corresponding decoding and precoding operation.S-LMMSE channel estimation algorithm requires the inverse operation of matrix,and the complexity of calculation will rise sharply with the increase of deployed antennas.Therefore,EW-LMMSE and LS channel estimation algorithm applied in multi-antenna systems is studied in this thesis.EW-LMMSE only uses the elements on the main diagonal of the spatial correlation matrix to perform channel estimation.While LS channel estimation does not require any matrix inversion,which is very helpful in reducing the estimation errors that caused by using partial statistics when users move at a high speed.Large scale fading decoding is wildly used in cellular network system,which maximize the user's achievable rate with a large amount of statistical data calculation.While traditional centralized decoding can be implemented in access points,which is helpful in reducing the calculation amount but with poor system performance.Therefore,this thesis proposes a centralized decoding method based on large-scale fading coefficients.Simulation results show that this decoding method improves achievable rate comparing with traditional centralized decoding.For lower backhaul requirement and higner user achievable rate,received uplink signals processed at CPU in user-centric cell-free massive MIMO is proposed in this thesis.At the same time,the estimated channel statistics needed in MMSE combining vector is also significantly reduced in this system.In order to improve the system performance,MMSE combining vector is proposed in this thesis for signal processing in user-centric cell-free massive MIMO system.The power control strategy based on user allocation scheme in user-centric cell-free massive MIMO system is also studied in this thesis.Max-min power control strategy applied widely in cellfree massive MIMO system,in this thesis it is proposed in user-centric cell-free massive MIMO and provides uniformly good service to all users,regardless user's transmission environment.User assignment strategy is performed mostly according to some kind of channel statistics.While the most ideal strategy is by maximize the user's spectral efficiency,while this strategy can be replaced by selecting the access points that maximize some kind of approximate closed-form SINR.Therefore,this thesis proposes that users can select access points according to the energy of signals sent by all access points.Simulation result demonstrates that the proposed user allocation scheme makes the system performance improved.Since this algorithm is performed on the user side,there is no need to reassign the access point,so the algorithm is optimized.