A Manifold Learning Based User Clustering and Beamforming Design in MIMO-SCMA Systems

In this paper, we study the user clustering and down-link beamforming problem in multiple-input multiple-output sparse code multiple access (MIMO-SCMA) systems. A K-means user clustering method based on manifold learning is proposed, which combines SCMA technology to limit the number of users within each cluster. Subsequently, a beamforming method is developed by the block diagonalization (BD) technology to mitigate inter-cluster interference, and different clusters share a common codebook set of SCMA. Within each cluster, individual users employ distinct codebooks consisting of sparse codewords to effectively minimize intra-cluster interference. Simulation results demonstrate the superior performance of the proposed method.