Precoding Methods For Massive MIMO Downlink

To meet the constantly increasing demands for higher wireless data rates,massive multi-input multi-output(MIMO)technology has drawn considerable attention.By employing a large number of antennas at the base station to serve multiple user equipments at the same time-frequent slot simultaneously,massive MIMO can improve spectral efficiency and link reliability.Precoding is the key technology in downlink transmission for non-cooperative multiple user MIMO systems.In this thesis,precoding technology for massive MIMO systems will be investigated.Firstly,several traditional precoding methods are investigated,and then,a MMSE precoding method based on eigenvalue decomposition of user channels is proposed.The performance of traditional precoding methods are limited by the correlation of user channels and will decreases when the user channels are strongly correlated.The MMSE precoding based on the pre-process of eliminating the channel correlation can increase the performance of MMSE precoding immensely.Simulation results show that the MMSE precoding based on pre-process of eliminating channel correlation outperform traditional precoding methods.Next,a user scheduling and beam allocation algorithm for massive MIMO is proposed.In massive MI-MO FDD systems,the cost of channel estimation and feedback increases linearly with the number of antennas at the base station,which makes accurate channel estimation very difficult when channel correlated time is limited.A two stage precoding method is proposed based on the sparsity of beam domain channel:the first stage aims to user scheduling and beam allocation based on statistical channel state information;the second stage precoding aims to eliminate the interference between user by RZF precoding based on instantaneous CSI.Simulation results show that the greedy algorithm not only reduces the cost of channel estimation by compressing the dimension of channel,but also has a better performance with reasonable computation com-plexity.Finally,a union user scheduling and RF chains allocation algorithm for mm-wave massive MIMO sys-tems is proposed.In mm-wave systems,user equipments can also be equipped with large number of antennas because of the short wave length of mm-wave.Besides,the characteristic of high attenuation of mm-wave results in the sparsity of RF chain between both the base station and user equipments.User scheduling and RF chains allocation at both BS and UE sides based on the sparsity of RF chains can not only reduce the cost of channel estimation by compressing the dimension of channel,but also increase the system performance by improving the power efficiency.In the algorithm,we first choose the users and RF chains at the base station side under the criterion of maximizing sum rates.To this end,we transform the problem of sum rate maximization into a difference of convex function programming,and then,solve the problem using convex optimization theory.After determining the RF chains at the BS side,we use a greedy method to choose the RF chains at UE side.Simulation results show that this algorithm reduces the cost of channel estimation and improves the system performance.