Statistical Power Allocation for Downlink Two-User Power-Domain MIMO-NOMA with Excess Degrees of Freedom

Abstract:
In this paper, we investigate statistical power allocation (SPA) for downlink two-user power-domain multipleinput multiple-output (MIMO) non-orthogonal multiple access (NOMA) to enhance the ergodic rate performance. We utilize a modified generalized singular value decomposition (M-GSVD) based precoding scheme which simultaneously diagonalizes the MIMO-NOMA channels of the users into multiple single-input single-output (SISO) NOMA channels, thereby reducing the decoding complexity. The M-GSVD precoding scheme is also able to exploit excess degrees of freedom at the base station to improve MIMO-NOMA performance. Based on finite-size random matrix theory, we derive novel expressions for the ergodic achievable rate of MIMO-NOMA with M-GSVD precoding and non-equal power allocation, which are then used for SPA and for obtaining the ergodic achievable rate region. The derived expressions are validated with Monte-Carlo simulations. Our results show that the proposed scheme outperforms M-GSVD precoding with equal power allocation and orthogonal multiple access, thereby demonstrating the benefits of SPA.