Wireless Powered Communication With Finite-alphabet Inputs

Wireless powered communication(WPC)is a promising research area which integrates microwave power transfer(WPT)and wireless communication.Replacing batteries of the nodes may be inconvenient or even impossible in some wireless networks such as a large scale wireless sensor network.The power beacon can provide an attractive solution by powering the nodes with dedicated radio-frequency(RF)based energy transfer.Thus,WPC is regarded as a promising technology to provide substantial wireless energy for wireless networks.Multipleinput multiple-output(MIMO)is a key technique of WPC,which has the potential to significantly improve the spectrum efficiency and provide remarkable performance gain.However,a considerable performance loss usually occurs when the design based on Gaussian inputs is used for systems with practical finite-alphabet inputs.This thesis is based on the precoder design for practical MIMO WPC systems.The main contributions are as follows:Firstly,two low-complexity precoded modulation methods,One-Step Precoding and Qucik Precoding,with practical inputs is proposed,which utilize modulation diversity technique.Simulation results shows that the proposed methods reduces computational burden with a little performance loss.In the following,the precoder designs are considered in three MIMO WPC sceneries,where the proposed methods are applied as well.Secondly,we study the simultaneous wireless information and power transfer(SWIPT)systems with finite-alphabet inputs.A precoding matrix design for MIMO SWIPT systems with practical inputs is considered.The design can be formulated as maximizing the mutual information given the energy level and the power constraint for achieving a so-called rateenergy region.The formulated design is non-concave,which can be reduced to a convex power allocation problem of the precoding matrix and solved by the gradient-descend method.Several rate-energy regions can be achieved by trading off between the information rate and the harvested energy with different schemes.Numerical examples indicate that the proposed scheme with finite-alphabet inputs provides significant performance gain over existing schemes with Gaussian inputs.Thirdly,another typical WPC scenery,wireless powered communication network(WPCN),with practical inputs is discussed,where a hybrid access point(H-AP)replenishes users by wireless energy transfer(WET)in the downlink,and the users utilize the harvested energy for wireless information transfer(WIT)through multiple access channel(MAC)in the uplink.A three dimension algorithm that jointly optimizes the downlink energy precoders,uplink information precoders,and time allocation between the downlink and uplink phases is proposed for maximizing the uplink sum-rate.Given time allocation,we first adopt energy beamforming and obtain information precoders by applying the gradient descent method.Then,the time allocation can be solved by the golden section search method.Simulation results illustrate the significant performance gain compared to existing methods.Finally,from the standpoint of finite-alphabet inputs,a specific scenery of WPCN,called wireless powered sensor network(WPSN),is considered for the sum throughput maximization problem.The power beacon performs WET to users,then each user makes use of received energy for WIT to the information sink by time division multiple access(TDMA).We propose a two dimension method to jointly allocate power and time.Firstly,given the time allocation,the non-concave precoder design is reduced to a convex power allocation problem,which can be efficiently solved by the gradient decsent method.Then,a semi-closed form solution for the suboptimal time allocation is obtained.Compared to existing methods,numerical examples show that the proposed algorithm offers significant performance gain.