Power Allocation Optimization And Performance Analysis In Non-orthogonal Multiple Access Schemes

Fundamental challenges in future wireless networks include ultra-high traffic density,massive connectivity,extremely high data rates,etc.Nonorthogonal multiple access technology has been recognized as one of the promising candidates to meet the diverse demands for the fifth generation(5G)wireless networks.In conventional orthogonal multiple access(OMA)systems,one user is normally allocated a single radio resource,such as time and frequency.In contrast,nonorthogonal multiple access technologies encourage multiple users to share a same radio resource,which can effectively increase the number of active users and system capacity,as well as reducing device power consumption.As a new multiplexing resource,power domain provides an additional dimension for designing multiple access.As a consequence,it is an important issue to increase the spectral efficiency and user fairness of nonorthogonal multiple access systems by designing power allocation strategies carefully.In this thesis,we focus on the optimization design of power allocation algorithms,together with the analysis of related theoretical issues and system performance criteria.Our investigations shall be conducted for both perfect and imperfect channel state information(CSI),in various scenarios including singleinput and single-output(SISO)systems,multiple-input and multiple-output(MIMO)systems,millimeter Wave(mmWave)systems,and multicell cooperative systems.In general,the existing research works focus on the investigation of power domain based nonorthogonal multiple access(NOMA)system with perfect CSI and fixed NOMA configurations,such as fixed power allocation coefficients and fixed decoding order.Due to the limited system overhead,the perfect CSI is hard to available at the base station(BS).Therefore,an outage event may happen in NOMA systems,which will decrease the potential benefits of NOMA.In this thesis,we consider a downlink SISO-NOMA system,where only the second order statistics of users' channels are available at the BS.We first derive the optimal decoding order of NOMA in terms of the outage probability.Then,we develop the optimal power allocation schemes in closed-form to minimize the transmit power and maximize user fairness for NOMA systems,respectively.Furthermore,the power difference between NOMA systems and OMA systems under outage constraints is attained.Finally,the provided simulation results demonstrate that NOMA outperforms OMA with the two proposed power allocation schemes.In addition,in contrast to the Gaussian-input signal,we consider the multi-user SCMA downlink system with finite-alphabet input.Then,we a novel power allocation scheme is developed to enhance user fairness.Simulation results demonstrate that the proposed power allocation algorithm is capable of enhancing the performance significantly compared to the equal power allocation scheme.The conventional MIMO-NOMA systems are generally investigated with fixed system configuration,such as perfect CSI and fixed the number of user clusters.In this thesis,weinvestigate the optimization of the utility function for the MIMO-NOMA system with multiple user clusters by considering two types of imperfect CSI.For the optimization problem under CDI,this thesis first derives the outage probability expressions of users in closed-form.Then,an efficient successive convex approximation(SCA)algorithm based on first-order approximation and semidefinite programming(SDP)is developed.For the optimization problem under channel estimation uncertainty,we approximate the outage probability expression by central limit theorem(CLT).Moreover,an iterative algorithm for the joint power allocation and receive beamforming design is developed to maximize the system utility.In addition,the convergence and the feasibility are discussed for the two formulated problems.The presented simulation results validate that the proposed two algorithms outperform MIMO-OMA and MIMO fixed NOMA(MIMO F-NOMA)with fixed power allocation and beamforming.Furthermore,we also investigate the power minimization problem for MIMO-NOMA systems.In particular,the algorithms for joint power allocation and receive beamforming designs under perfect and imperfect CSI are developed,respectively.To enhance the benefits of NOMA in mmWave spectrum,we consider a downlink mmWaveNOMA system by optimizing the designs of user scheduling and power allocation.The branch and bound(BB)approach is invoked to obtain the optimal power allocation policy.The generated optimal user scheduling and power allocation solution was served as an upper bound of the sum rate.To elaborate further,a low complexity suboptimal approach is developed for striking a good computational complexity-optimality tradeoff,where matching theory and SCA techniques are invoked for tackling the user scheduling and power allocation problems,respectively.Moreover,the analysis of the convergence and the computational complexity for the proposed algorithms are provided.Simulation results reveal that: i)the proposed low complexity solution achieves a near-optimal performance;and ii)the proposed mmWave NOMA system is capable of outperforming conventional mmWave orthogonal multiple access(mmWave-OMA)systems in terms of sum rate and the number of served users.The combinational features of user clustering impose challenges on solving the sum rate optimization of mmWave-NOMA systems.In this thesis,we propose a new machine learning based framework for mmWave-NOMA systems,where the users follow an spatially clustering distribution model.In particular,we develop a K-means based machine learning algorithm for user clustering by efficiently exploiting the feature of NOMA transmission.Moreover,for a practical dynamic scenario where the new users keep arriving in a continuous method,we propose a K-means based on-line user clustering algorithm to reduce the computational complexity.In addition,to further enhance the performance of the proposed mmWave-NOMA system,we derive the optimal power allocation policy in a closed form by using the successive decoding feature.Simulation results reveal that: i)the proposed machine learning framework achieves a near optimal performance of mmWave-NOMA systems by comparing it to the conventional user clustering algorithms,and thus is validated in practical mmWave-NOMA scenarios;ii)the proposed K-means based on-line user clustering algorithm provides a comparable performance to the convention K-means algorithm and strike a good balance between performance and computational complexity.Quality of experience(QoE)is an important indicator in the 5G wireless communication systems.In this thesis,we investigate the optimization of QoE for multi-cell multicarrier nonorthogonal multiple access(MC-NOMA)networks.An optimization problem is formulated with the objective of maximizing the weighted sum mean opinion score(MOS)of users in the networks.To solve the challenging mixed integer programming problem,we first decompose it into two subproblems,which are characterized by combinational variables and continuous variables,respectively.For the combinational subproblem,a three-dimensional(3D)matching problem is proposed for modelling the relation among users,BSs and subchannels.A twostep approach is proposed to attain a suboptimal solution.For the continuous power allocation subproblem,the BB approach is invoked to obtain the optimal solution.Furthermore,a low complexity suboptimal approach based on SCA techniques is developed for striking a good computational complexity-optimality tradeoff.Simulation results reveal that the proposed NOMA network outperforms conventional OMA networks in terms of QoE,and the proposed algorithms for sum-MOS maximization can achieve significant fairness improvement against the sum-rate maximization scheme.