As an enabling technology for emerging and future generations of wireless networks, sparse code multiple access (SCMA) offers major improvements in terms of spectral efficiency and massive connectivity. Although the message passing algorithm (MPA) for SCMA decoding at the receiver side can achieve near optimum performance, it entails high computational complexity. In this paper, to address this issue, 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, residual blocks are employed to tackle the problems of accuracy saturation and vanishing gradients with deep learning based decoder, while batch normalization is utilized to enhance the stability and robustness of the decoder. 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.

In this paper, we investigate the application of downlink beamforming along with non-orthogonal multiple access (NOMA) in a multi-user multiple-input multiple-output (MIMO) system. The joint optimization framework for user clustering, downlink beamforming and power allocation scheme is formulated as a novel mixed-integer non-linear program (MINLP), where the aim is to minimize the total transmission power while satisfying quality-of-service (QoS), user clustering and power constraints. Owing to the non-convexity and combinatorial nature of the problem, obtaining an optimal solution is challenging. To tackle this issue, we first develop an algorithm based on branch-and-bound (BB), whereby the feasible space is successively partitioned and searched by means of lower and upper bounds on the objective function. While this algorithm is shown to return an$\epsilon $-optimal solution within a finite number of iterations, it entails high computational complexity. Considering this limitation, we then reformulate the original problem into a more tractable form and conceive a low-complexity algorithm for its solution based on the penalty dual-decomposition technique. The proposed joint design algorithms for MIMO NOMA are evaluated by means of simulations over mmWave channels. Results show significant improvements in terms of total transmit power and spectral efficiency compared to benchmark approaches.

In this paper, we study the application of spatial user clustering along with downlink beamforming in multiple-input multiple-output sparse code multiple access (MIMO-SCMA) systems. A user clustering algorithm based on a constrained K-means method is proposed to limit the number of users in each cluster. Subsequently, a two-stage beamforming approach is developed in which a cluster beamformer and user-specific beamformer obtained from each stage are combined to form the final beamformer for each user. Specifically, in the first stage, the block diagonalization technique is employed to design cluster beamformers so that the inter-cluster interference is removed. In the second stage, an optimization problem is formulated to determine user-specific beamformers for all the users such that the total transmit power is minimized under signal-to-interference-plus-noise ratio (SINR) 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 scheme in terms of transmit power, and spectral efficiency over benchmark approaches.