Wideband Linear Precoder Design Under Per-Antenna Power Constraints

Multiple-input-multiple-output(MIMO)technique has been widely used to boost spectral efficiency in modern wireless communication systems.Under multiuser MIMO downlink scenarios,higher throughput can be achieved by the precoding technique,which optimizes the transmission strategy by utilizing knowledge of the channel state information at the transmitter.Despite the mature precoding schemes in current wireless standards,the use of large antenna arrays and large bandwidth poses new challenges for multiuser MIMO precoding in 5G and future mobile communication systems,such as the lack of efficient algorithms for solving the linear precoder subject to per-antenna power constraint(PAPC),the overwhelming multi-carrier complexity of precoding on a subcarrier basis,and the inevitable rate loss of precoding with the Shannon capacity objective in practical systems.In view of the above challenges,this dissertation studies low-complexity wideband linear precoding algorithms for downlink multiuser MIMO systems,with a comprehensive consideration of several limiting factors in precoding design such as the frequency selective fading channel,PAPC,and adaptive modulation and coding(AMC).The main work can be summarized as follows.1)The design of wideband precoding for downlink multiuser MIMO systems is considered,which aims to maximize the Shannon capacity under PAPC.Firstly,the wideband precoding strategy is proposed,where all subcarriers in a resource block group(RBG)share one wideband precoder,to handle the overwhelming complexity of precoding on a subcarrier basis for systems with abundant subcarriers.Then,taking Shannon capacity as the objective,the optimal wideband precoder and the suboptimal wideband precoder are designed,respectively.The optimal wideband precoder corresponds to one stationary solution to the wideband weighted sum rate maximization(WSRMax)problem under PAPC.Two iterative algorithms are proposed for solving this problem.The first one is developed from the existing double-loop algorithm and improves the solving of the subproblem at each iteration.The second one is a new single-loop algorithm exploiting the inherent separability of the feasible set under PAPC,in which each updating step is performed simply with closed-form expressions.In addition,by introducing a low-complexity near-optimal zero-forcing(ZF)algorithm that satisfies PAPC into wideband precoding design,the suboptimal wideband precoder is obtained,which can approach the performance achieved by the optimal wideband precoder at reduced complexity under some scenarios.2)The design of wideband precoding for downlink multiuser MIMO systems is considered,which aims to maximize the AMC throughput under PAPC.As a real communication system supports only a finite number of modulation and coding schemes(MCSs),applying precoders with the Shannon capacity objective to practical systems can result in a loss in spectral efficiency.To solve this situation,this work considers wireless systems that employ adaptive modulation and coding(AMC),and studies the modeling and optimization with wideband precoding of the throughput.Firstly,an effective signal-to-interference-plus-noise ratio(SINR)mapping method based on generalized mutual information(GMI)is put forward,which establishes the functional relationship between the SINRs on different subcarriers and spatial streams and the resulting block error rate(BLER)after decoding in multicarrier and multistream transmission.Based on the BLER function,a functional relationship is also formulated between the user’s throughput and the SINRs of that user in a wireless system employing AMC.Finally,taking AMC throughput as the objective,the optimal wideband precoder under PAPC is proposed.3)For downlink three-dimensional(3D)massive MIMO systems,the design of two-dimensional wideband precoding that satisfies PAPC is considered.This work focuses on massive MIMO systems equipped with a uniformly rectangular antenna array(URA)at the base station and studies low-complexity wideband precoding algorithms by exploiting the Kronecker structure of 3D MIMO channel.Firstly,assuming that the angular spread(AS)in elevation is zero such that 3D MIMO channel admits a perfect Kronecker product structure,an efficient two-dimensional wideband precoding scheme that meets PAPC is proposed.The idea is to divide the full precoding matrix into two smaller matrices,one of which is to suppress multiuser interference,while the other is to enhance the desired signal for each user.Their respective solving methods are also offered.Compared to conventional full precoding,two-dimensional precoding has a much lower complexity due to the reduction in the optimization problem size,and an eligible spectral efficiency thanks to its utilization of the channel structure.Next,by performing the Kronecker decomposition of the channel matrix,the proposed two-dimensional precoding scheme is extended to the general scenario where the AS in elevation is not negligible.Meanwhile,the impact of nonzero AS on the performance of two-dimensional precoding is reduced by improving the solving of the interference suppression matrix.