SWIPT-Enabled Full-Duplex NOMA Networks With Full and Partial CSI

This paper investigates the simultaneous wireless information and power transfer (SWIPT)-enabled cooperative non-orthogonal multiple access (NOMA) with full-duplex (FD) relaying. Two optimization problems are formulated to minimize the required transmit power with both full channel state information (CSI) and partial CSI (i.e., channel distribution information (CDI)). As both problems are non-convex without explicit solution methods, the semidefinite relaxation (SDR) is applied to relax them firstly. Then, for the full CSI case, a bilevel programming (BLP)-based method is presented to find the global optimal solution, and as an alternative, a successive convex approximation (SCA)-based method with arithmetic and geometric means (AGM) inequality is designed to achieve the suboptimal solution with low complexity. Moreover, in a special case with single transmit antenna, the closed-form expression of the optimal power allocation is derived for a given power splitting (PS) ratio and then the global optimal solution is found by using Bisection method. For the CDI case, the first-order Taylor expansion is adopted to approximate the non-convex constraints and then a SCA-based algorithm is designed to find the suboptimal solution. Numerical results validate our obtained theoretical results and proposed solution methods, and also show that by migrating FD relaying into SWIPT-enabled cooperative NOMA, a considerable system performance gain can be achieved for both full CSI and CDI cases. Besides, we also discuss the gain brought by NOMA and FD separately and the effect of the relay location on the system performance. It shows that FD may bring more contribution to the system performance improvement than NOMA.