Research On Precoding Technology For Dense Distributed Antenna Systems

In modern communications,the rapid increase of user requests has made research of 5th generation mobile communication on a fast speed.The traditional centralized antenna systems cannot afford the user requests nowadays any more.At the same time,the new form of dense distributed antenna systems have the advantage of lower average access distance and lower request of power on antenna units.The signal converage in the cell has also been enhanced by distributed antenna systems,so that the transmit power of the systems can be under well control.This thesis is based on the background of distributed antenna systems.It first researches on the new structure in distributed antenna system which is called cloud radio access network.Beamforming vector design and RAU selection are also discussed under C-RAN structure.Proposition Fairness is considered in this thesis to study on different user power allocation situations under different strategies as well.To decrease the power consumption of the whole system,different RAU status is considered under MIMO-OFDM network in the thesis.An algorithm for RAU status control is presented in the thesis.Last but not least,RAU cache is considered under multicast transmit strategy which is used to harness RAU cache at the best.An algorithm is presented in the thesis,where different cache modes are compared to analyse their performance.Firstly,we introduce path loss model,shadow fading model and small-scale model.Then,we introduce the structure and features of distributed antenna systems.The capacity of distributed antenna systems are discussed through formulas.We also introduce two basic structures of cloud radio access network and their advantages and disadvantages.We discuss the influence of RAU power constraint and backhaul power constraint on cloud radio access system capacity through simulations.Secondly,we consider network backhaul power constraint and RAU power constraint under C-RAN to design beamforming vectors and dynamic clustering for users.We present a weighed sum rate maximization problem in this thesis where the target variables are more than one.As a result,we use the block coordinate descent method to solve the problem by steps.The whole algorithm formed uses sparse beamforming to design beamforming vectors to solve user clustering and beamforming in just one step.We also add proportional fairness when using weighed sum rate maximization.We compare proportional fairness,random beamforming and round-robin through simulation to consider their influence on user performance.In addition,we use RAU with changeable status under C-RAN to form a system power consumption minimization problem.We consider RAU with two different modes including active and sleep.We also consider RAU has different basic power consumption under these two modes.We consider this problem under MIMO-OFDM,where balance between different subcarriers is considered through energy efficiency.We design an iteration algorithm which searches the maximum number of RAU that can be turned into sleep mode to reach the aim of minimizing the power consumption of the system.We use simulation to analyse the performance of the algorithm.Comparison is made between the algorithm we propose and the algorithm without RAU status changing,which leads to the conclusion that the algorithm we propose can actually reduce the power consumption of the system.Last but not least,we use RAU cache under C-RAN to form a system power consumption minimization problem aiming at user clustering and beamformer designing.We use multicast to form user clusters,where users are divided through the contents they requested.With two variables under optimize,we first propose an algorithm using exhausting search to find the multicast groups each RAU serves.Then the power consumptions under different circumstance are compared to find the minimum power consumption clustering.Except this method,we solve the problem through transforming the target function into its estimation through sparse beamformer designing,where the former question can be turned into a convex problem.We also discuss the performance difference between different cache strategies through simulation.