The Design Of Pilot And Precoding Based On Nonconvex Optimization

With the rapid development of mobile communication technologies,the devices connected to the wireless network have explosive population growth.The wireless communication systems of super high-speed,low latency,high reliability and high energy efficiency is an important foundation to meet the requirement of lots of emerging communications services.The channel estimation,symbol detection and precoding are the key technologies for the system with high reliability and high energy efficiency.However,the approximate optimization methods in current research works are not suitable and with high computational complexity.This thesis focuses on the research of channel estimation and symbol detection of orthogonal frequency division multiplexing(OFDM)and energy-efficient precoding of multiple antennas.The effective optimization schemes have been proposed,where the nonconvex objective functions and nonconvex constraints have been transformed in the optimization of pilot and precoding.We develop the iterative algorithms,which quickly converge at least to the locally optimal solutions.This thesis focuses on the above problems and the main contributions of this thesis are listed as follows.1.We consider the power allocation of pilot symbol for OFDM systems with time-varying channel.Based on the system model,we have proposed two optimization schemes,where the effective signal-to-interference-plus-noise ratio(SINR)after data symbol detection and the sum rate of data transmission are set as the objective functions,respectively.Since the effective SINR and sum rate based on linear minimum mean-square error(LMMSE)channel estimation are nonconvex function,we derive the lower bound of the effective SINR and sum rate,which are chosen as the objective functions,respectively.Nevertheless,we shall show that the solution to the problem of maximizing the lower bound of effective SINR admits a closed form,while the solution to the problem of maximizing the lower bound of sum rate can be globally located by our iterative algorithm.In terms of channel estimation,we derive the construction of whole path gains from the average path gains within a few consecutive OFDM symbols based on the framework of basis expansion model(BEM),which reduces the complexity of channel estimation.2.We consider the pilot sequence design for OFDM systems with time-varying channel.The design of pilot sequence to minimize the mean squared error(MSE)of the channel estimate is proposed under a LMMSE estimator for average path gains.Due to the existence of interferences of pilot and data subcarriers,the MSE of the LMMSE estimator is nonconvex function.Hence,the design of pilot sequence becomes a very difficult optimization problem.We use the successive quadratic programming(SQP)to transform the MSE of the LMMSE estimator to concave quadratic function and develop a path-following optimization procedure,which improves the MSE in every iteration and quickly converges at least to its local optimal solution.The developed path-following procedure can also be adapted to design pilot sequences for the least-square(LS)and maximum-likelihood(ML)estimators.Extensive simulation results demonstrate that in the low-SNR regime the MSE of our pilot sequence is lower 4d B than the current work.3.We consider the pilot sequence design for multiple input multiple output(MIMO)-OFDM systems.The objective for designing pilot sequence is to minimize the entropy of the error between channel state and its minimum mean square error(MMSE)estimator.Unlike the case with full length training sequences,where the pilot sequence design is formulated as a semi-definite programming(SDP)with the available computational solution,the problem of short-length pilot sequence design is highly nonconvex and thus is very computationally challenging,where the rate occupancy of the pilot sequence within one OFDM symbol is just 1/128.We use the SQP to transform the objective function into convex quadratic function and propose an iterative procedure to solve it.4.In terms of design of energy efficiency(EE)precoding for multicell network with base stations(BSs)in full duplex model,we have proposed the path-following algorithm with convex quadratic program for optimizing EE.In the multicell network,uplink and downlink users are equipped with multiple antennas and operate in a half-duplex mode.Given that interference is very severe,the downlink and uplink throughputs are nonconcave,and the quality of service(Qo S)in terms of user’s minimum rate constitutes difficult nonconvex constraints.Therefore,the optimization for EE performance is very difficult.We propose to transform the downlink and uplink throughputs to concave quadratic function and also transform Qo S constraints to convex quadratic constraints.Then,we develop a low-complexity path-following algorithm that only invokes one simple convex quadratic program at each iteration.The simulation results demonstrate that the proposed algorithm has the low computational complexity,and in the low self-interference regime the EE of full duplex is double of that of half duplex.