Design of user clustering and precoding for downlink non-orthogonal multiple access (NOMA)

In this paper, we propose a novel user clustering algorithm for non-orthogonal multiple access (NOMA) considering the channel correlation between users and the channel gain. We also adopt sum-rate maximization approach to find an optimal precoding matrix in the multi-user multiple-input-multiple-output (MU-MIMO) setup after the user clustering. Grouping two users in a single beam to serve users in a non-orthogonal fashion raises multi-user interference (MUI). To tackle this problem, the clustering algorithm considers the channel correlation to select users aligned with same direction. Furthermore, the channel gain is also reflected in the clustering algorithm to enhance signal-to-interference-plus-noise ratio (SINR) inspired by the fact that the optimal precoding matrix is less sensitive to channel direction. Optimization algorithm over the precoding matrix is proposed based on Majorization Minimization (MM) approach for a non-convex objective function. We demonstrate the superiority of the proposed schemes comparing with conventional ones through numerical results.