Throughput Maximization in Grant-Free Power-Domain NOMA for 3D Distributed Users by Stochastic Geometry

This paper maximizes the throughput in the grant-free power-domain non-orthogonal multiple access (GF-NOMA) for users distributed in a 3D space by optimizing the power level selection strategy. To maximize the throughput, we propose a stochastic geometry-based analytical model for an offered load per level (called per-level offered load), deciding the throughput, for 3D distributed users. The key idea is to reflect line-of-sight conditions depending on the altitudes of such users to power level constraints. This key idea enables our model to stochastically limit selectable power levels along with LoS conditions and derive a selection strategy suitable for offloading an inhomogeneous per-level offered load caused by LoS conditions to maximize the throughput. Our optimization results showed that the derived power level selection strategy achieved 16% higher throughput than the baseline strategy, where each user randomly selects a power level under the power level constraints.