Robust Precoding For Massive MIMO Mobile Communications

With scarce spectrum resources,the emerging and constantly changing services prompt an increasing hunger for more efficient and reliable transmission techniques in the information society.Massive multiple-input multiple-output(MIMO)transmission is the crucial physicallayer wireless technique of 5G(5 generation)and will be further developed and applied in the future 5G evolution and 6G(6 generation).Different realizations of massive MIMO systems require distinct transmission designs.The precoder designed for three-dimensional(3D)massive MIMO systems is required to effectively exploit the channel’s degrees of freedom in the elevation to improve the system throughput.In multi-cell scenarios,coordinate beamforming among cooperating base stations(BSs)is expected to suppress inter-cell interference substantially.At the same time,joint precoder design for network MIMO systems is believed to improve spectral efficiency further.Nevertheless,regardless of specific realizations,in typical mobile scenarios,imperfect channel state information(CSI)affected by channel aging is the bottleneck that restricts the performance of massive MIMO systems.In this dissertation,the robust precoder design is investigated with imperfect CSI.Firstly,the robust downlink precoding for three-dimensional(3D)massive multi-input multi-output(MIMO)configuration with matrix manifold optimization is investigated.With the a posteriori channel model,the robust precoder design is formulated to maximize an upper bound of weighted ergodic sum-rate under a total power budget.The generalized eigenvector structure for optimal precoder is derived with matrix manifold optimization.Since the precoding of multiple users is coupled in the structure,the objective function is maximized for each user in alternation.The solution to each problem is the generalized eigenvector corresponding to the maximum generalized eigenvalue is revealed.Following this,an iterative algorithm is proposed,followed by its convergence analysis.Furthermore,a Riemannian conjugate gradient(RCG)method is proposed to solve the generalized eigenvalue problem(GEP)for higher efficiency in the precoder design algorithm.Secondly,the robust precoding for coordinated multi-cell massive MIMO systems with imperfect CSI formulated as the a posteriori channel model is investigated.A weighted ergodic sum-rate maximization problem is formulated for optimal robust precoder design against multicell interfering downlink channels.Within the minorization-maximization(MM)method framework,an iterative precoder structure,which is shown to generalize the well-known weighted minimum mean square error(WMMSE)precoder with the a posteriori channel model,is developed.Further,a low complexity robust precoder design using the deterministic equivalents is proposed to lighten the computation burden.A design framework for decentralized processing is provided to facilitate the coordinated precoder implemented in practice.Based on the design framework,three specific user-transparent decentralized processing strategies with distinct information exchange requirements and computation costs are devised.Simulation results confirm the performance superiority of the proposed precoders over the non-robust precoder and non-cooperated robust precoder.Thirdly,aiming at maximizing the weighted ergodic sum rate,the robust precoding for network massive MIMO systems with the a posteriori channel model is investigated.The transmission of network massive MIMO system is classified into full service(FS)and partial service(PS)strategies.The precoder design problem is formulated with the matrix manifold optimization for the FS strategy,and the optimality condition for the precoder design is presented.With the MM method on Riemannian manifolds,an iterative precoder design that can converge to the optimal point on the Riemannian manifold is proposed.The iterative precoder design possesses a coupled structure that can not be computed directly.Based on the Riemannian block gradient descent(RBCD)method,the precoders of different BSs are designed in alternation,endowing the iterative precoder design with a two-level iteration structure.For the PS strategy,the problem is established as the one for the FS strategy constrained by more power constraints and further transformed into an optimization on the intersection of two Riemannian submanifolds.The relation between the structure of the optimal precoder for the PS strategy and the one for the FS strategy is revealed.Then,the robust precoder design for the FS strategy is generalized to the PS strategy.Simulation results show that the proposed robust precoder design’s sum-rate performance for network massive MIMO systems is superior to the non-cooperative robust precoder.