MIMO Precoding With Finite-alphabet Inputs

The unprecedented growth of mobile terminals and data traffic stimulates the development of wireless communication technology.By 2019,the global mobile data traffic is expected to grow to 24.3 exabytes per month.It becomes evident that the current 4G network will not be able to deal with these emerging demands,thus deploying 5G communication network is imperative in the future.In order to achieve significant improvement on performance and efficiency,5G network incorporates three basic technology: millimeter wave,massive multiple-input multiple output(MIMO),and heterogeneous network.In addition,the physical layer security may also be applied to ensure secure communication in 5G network.This thesis is based on the design of MIMO precoding for physical layer security and millimeter wave communication.Its main contributions are listed as follows:Firstly,we investigate a new class of optimization,which provides a unified framework for linear precoding in single/multi-user multiple-input multiple-output(MIMO)channels with arbitrary input distributions.The new optimization is called generalized quadratic matrix programming(GQMP).The objective function and constraints of GQMP problems are nonconvex,thus GQMP belongs to the class of nonconvex optimization.Then we design an efficient algorithm to solve GQMP problems,and the proposed algorithm is guaranteed to converge to a Karush-Kuhn-Tucker(KKT)point.Finally,we apply GQMP to a downlink underlay secure cognitive radio network.Based on the GQMP algorithm,we can design linear precoder to maximize the average secrecy sum rate of the network.Simulation results are provided to show the efficacy of our proposed algorithm from several aspects.Secondly,we design linear precoder to maximize the average secrecy rate sum rate for MIMO cognitive multiple access wiretap channel with finite-alphabet inputs and statistical channel state information(CSI).We propose a two layer optimization algorithm to search the optimal linear precoder.The main idea behind this algorithm is to reformulate the average secrecy sum rate as a difference of convex functions,and then generate a sequence of convex sets whose union can approach the non-convex feasible set.Subsequently,we maximize the average secrecy sum rate over each convex set by using the convex-concave procedure.Simulation results indicate that our proposed algorithm is superior to the conventional Gaussian precoding method in the medium and high signal-to-noise(SNR)ratio regime.Thirdly,we consider the fully-connected hybrid precoding design for millimeter wave(mmWave)MIMO systems.we present a sufficient condition and a necessary condition for hybrid precoding schemes to realize unconstrained optimal precoders exactly.Then we formulate the hybrid precoding problem as a matrix factorization problem,and show that the power constraint in our matrix factorization problem can be removed without loss of local and global optimality.Finally,we present closed form expressions for gradient and Hessian of the hybrid precoding problem.Then we utilize these information to design a hybrid precoding algorithm.Simulation results show that the proposed algorithm outperforms existing hybrid precoding algorithms.Finally,we consider a mmWave MIMO system with partially-connected hybrid precoding architecture,where each radio frequency(RF)chain is connected to a dynamic subset of antennas.We investigate the design of optimal connection scheme between RF chain and antennas using statistical CSI,and the design of hybrid precoding using either statistical CSI or mixed CSI.We first present a simple criterion to design the optimal connection between RF chain and antennas.Then we design a low complexity algorithm to solve the corresponding optimal connection problem.Then we derive a lower bound of the average mutual information for mmWave MIMO channels.The lower bound plus a constant shift serves as a very accurate approximation to the average mutual information,and its complexity is much lower than the original average mutual information.To further reduce the complexity,we also derive an accurate approximation of the proposed lower bound.Finally,we propose a manifold-based gradient ascent algorithm to design hybrid precoders.Simulation results demonstrate that the mixed CSI based hybrid precoding can approach the performance of instantaneous CSI based hybrid precoding.