PhD defence of Sara Norouzi - Non-Orthogonal Multiple Access for MIMO Wireless Communications

Abstract

Through spatial diversity, multiplexing or beamforming gain, the multiple-input multiple-output (MIMO) techniques can offer significant performance improvements in terms of user capacity, spectral efficiency, and peak data rates. Recently, the application of MIMO techniques along with non-orthogonal multiple access (NOMA) has aroused great interest as an enabling technology to meet the exacting demands of fifth generation (5G) and beyond 5G (B5G) wireless networks. In effect, by allowing multiple users to access overlapping time and frequency resources in the same spatial layer, NOMA has the potential to provide higher system throughput and solve the massive connectivity needed for future wireless networks. The primary objective of this thesis is to develop new approaches for multi-user MIMO NOMA systems from the perspectives of spectral and energy efficiency.

First, the joint design of user clustering, downlink beamforming and power allocation is formulated as a mixed-integer non-linear programming (MINLP) model for a MIMO NOMA system. In this problem the aim is to minimize the total transmission power while satisfying quality-of-service (QoS) and power constraints. To tackle this challenging problem, we reformulate it into a more tractable form and conceive two algorithms based on the branch-and-bound and penalty dual decomposition techniques for its solution. The performance of the proposed joint design algorithms for MIMO NOMA is validated by means of simulations over millimeter-wave (mmWave) channels. The results show the advantages of the proposed algorithms in terms of total transmit power and spectral efficiency over competing multiple access schemes.

Then, we study the application of spatial user clustering along with downlink beamforming for MIMO sparse code multiple access (SCMA) in a cloud radio access network (C-RAN). A user clustering algorithm based on a constrained K-means method is proposed to limit the number of users in each cluster. Subsequently, two iterative algorithms for beamforming design are developed by minimizing the total transmission power under QoS and fronthaul capacity constraints. The performance of the proposed user clustering and downlink beamforming approaches in MIMO SCMA systems is evaluated through simulations. The results provide useful insights into the advantages of the proposed schemes over benchmark approaches, in terms of transmit power and spectral efficiency.

Finally, we propose a novel SCMA decoder based on deep residual neural network (ResNet), wherein the decoder is trained to predict the transmit codewords. In our approach, batch normalization is utilized to enhance the stability and robustness of the decoder, while residual blocks are employed to tackle the problems with deep learning-based decoder such as accuracy saturation and vanishing gradients. The performance of the proposed ResNet decoder for SCMA is validated by means of simulations over AWGN and Rayleigh fading channels. The results show that besides a much-reduced complexity, the proposed decoder leads to improvements in term of bit error rate (BER) over competing deep neural network (DNN) based decoders.