Fractional Programming for Robust TX BF Design in Multi-User/Single-Carrier PD-NOMA (2001.00655v3)

Abstract: We present a new Beamforming-based (BB) Multiple-Input Single-Output (MISO)-Non-orthogonal Multiple Access (NOMA) scheme for Power Domain NOMA (PD-NOMA), in which the total transmit power consumption is minimized subjected to prescribed signal-to-interference-plus-noise ratio (SINR) requirements for each user, and under the assumption that only imperfect channel state information (CSI) is available at the transmitter. To this end, the fractional programming (FP)-based quadratic transform is employed to reformulate the non-convex SINR constraint of the original problem into a tractable quadratic form, which contains an estimate of the CSI error vector as a parameter. Taking advantage of the fact that the zero duality gap holds for the non-convex quadratic problems, a closed-form expression for an estimate of the CSI error vector is derived, completing the formulation. Finally, a novel iterative algorithm based on both the herein derived CSI error vector and the semidefinite relaxation (SDR) technique is contributed, which is shown to capable of efficiently solving the constrained min-power problem. Simulation results are given which illustrate the effectiveness of the proposed algorithm, which is found to sacrifice only small quantities of transmit power in return for a substantial increase in robustness against CSI imperfection.