A Novel Uplink NOMA Scheme Based on Low Density Superposition Modulation

Abstract

Comparing with traditional orthogonal multiple access scheme, non-orthogonal multiple access (NOMA) scheme can effectively meet the tighter requirements of the fifth generation (5G) wireless networks. Numerous NOMA schemes utilize the sparsity of the check matrix to get high performance with a moderate complexity. Sparse Code Multiple Access (SCMA) is a superior NOMA scheme, which bases on multi- dimensional sparse codebook. However, there exist only few works focusing on the design of low-density check matrix of SCMA. Short cycle maybe exist when the dimension of the SCMA check matrix is small. The existence of the short cycle will make a mistake value transmit back, which strengthens the possibility of misjudgment, and may degrade the decoding performance when using message passing algorithm (MPA). Besides, the regular check matrix of SCMA doesn't work well in ``near-far effect'' environments. In this paper, a novel uplink NOMA scheme named low density superposition modulation (LDSM) scheme is proposed to design an sparser and irregular check matrix, which can get diversity gain and coding gain and can improve not only the overall performance but also the edge users' performance. The SNR gains LDSM can get over SCMA is demonstrated in theory by the analysis of the average mutual information (AMI) of coded modulation (CM) systems in Rayleigh fading channels. Simulation results indicate that when power difference is 10dB, small dimensional LDSM and large dimensional LDSM can get 2dB and 2.5dB gains respectively over SCMA for overall performance. Both theoretic analysis and simulation results prove that the performance of LDSM outperforms that of SCMA.