Abstract
Nonorthogonal multiple access (NOMA) has been considered as a promising technology for fifth-generation wireless communication systems. In this paper, we consider a two-cell multiple-input-multiple-output NOMA downlink interference channel with one multiantenna base station (BS) and multiple multiantenna users in each cell. Each user treats interferences from adjacent cell as additional noises and does not attempt to decode them. This paper aims to design interference channel beamformers to maximize weighted sum capacity of the strongest users in two cells under quality of service requirements at other users and transmit power budget constraint at each BS. Such problem is complicated and nonconvex due to the expression of a logarithm of a determinant (\log \det) minus \log \det in the objective function and constraints. For the nonconvex problem, we put forward a successive convex approximation (SCA) based iterative algorithm. Since the computational complexity of SCA-based iterative algorithm is high, we also put forward an efficient majorization-minimization based iterative algorithm. Based on simulation results, we show that our developed NOMA schemes are superior to the zero-forcing-based NOMA, orthogonal-NOMA, and orthogonal multiple access schemes.