Research On Error Propagation Suppression In Non-Orthogonal Multiple Access

With the development of Internet of Things and cloud-based applications,there is an explosive demand of new services and data traffic for wireless communications.Therefore,the fifth generation(5G)communication systems propose higher requirements in data rate,low latency,and massive connectivity.In order to meet these high demands,non-orthogonal multiple access(NOMA)has been considered as one of the most potential technologies.The conventional orthogonal multiple access(OMA)technologies eliminate the multi-user interference by allocating orthogonal resources to different users and the spectrum efficiency is low.By using the superposition coding at transmitter and the successive interference cancellation(SIC)algorithm at receiver,NOMA allows the same frequency to be allocated to different users at the same time,which greatly improves the spectrum efficiency.SIC is the key technique of user detection in NOMA system.Nevertheless,a main drawback of SIC is the inter-user error propagation issue,because decision errors result in subtracting the wrong reconstructed signal from the composite multiuser signal.In this thesis,SIC algorithm in NOMA is studied from the perspective of improving the decision reliability of users detected previously.The main contributions of this thesis are summarized as follows:Firstly,for two-user NOMA downlink,a successive interference cancellation algorithm based on constellation rotation is proposed to improve the noise tolerance of transmitting signals.At the transmitter,when both users adopt BPSK or QPSK modulation,the base station allocates power according to the user channel gain.Then modulation signal of the user with poorer power is rotated by a certain angle,and is superimposed with the modulation signal of the user with stronger power.At the receiver,the SIC algorithm based on constellation rotation is used to detect the user with poorer power.In addition,an optimization problem is formulated by maximizing the minimum Euclidean distance of two-user superimposed constellation,and the optimal rotation angle is obtained by solving the problem.When both users adopt BPSK modulation,the BER performance of the proposed SIC detection algorithm is close to that of the perfect SIC.When both users adopt QPSK modulation,compared with imperfect SIC algorithm,the proposed SIC algorithm has a performance gain of about 2d B.Secondly,in order to obtain the optimal rotation angle,in the case of both users adopting MPSK modulation,an optimization problem is also established with the objective function of maximizing the minimum Euclidean distance of two-user superimposed constellation.When the power difference between two users is small or the modulation order is high,the superimposed constellation will intersect.When there is no intersection between the constellation points of the superimposed constellation,general condition for the optimal rotation angle is derived by using the symmetry of the constellation points and the periodicity of the minimum Euclidean distance.When there is intersection between constellation points of superimposed constellation,the SIC algorithm proposed in this thesis is no longer applicable.Finally,the SIC algorithm combined with channel coding is investigated.For the two-user NOMA downlink,the two users signals are superimposed after the convolutional channel coding,modulation and power allocation separately,and the SIC algorithm based on soft decision decoding is used at the receiver.Compared with the SIC algorithm based on hard decision decoding,the BER performance of the target user is improved by about 3d B.In addition,compared with traditional SIC detection without coding,the performance loss of SIC detection algorithm based on soft decision decoding is smaller,and therefore the reliability of detection of the target user can be improved by using channel coding.