Power-Level-Design-Aware Scalable Framework for Throughput Analysis of GF-NOMA in mMTC

Abstract

This article proposes a scalable framework to analyse the throughput of the grant-free power-domain nonorthogonal multiple access (GF-NOMA) and presents the achievable performance in the optimized offered load at each power level (called per-level offered load) by using our framework. Our analytical model reflects packet errors caused by power collisions, characterized by GF-NOMA, based on the power level design guaranteeing the required signal-to-interference-and-noise ratio (SINR). This key idea enables analyzing the throughput of a large-scale GF-NOMA system more accurately than the existing analytical models. Also, this key idea enables optimizing the per-level offered load rather than an uniform one in typical optimization problems related to the throughput: the throughput maximization or energy minimization problem with a throughput condition. Our analytical results highlight some key insights into designing future access control methods in GF-NOMA. First, our analytical model achieves an approximation error of only 0.4% for the exact throughput obtained by the exhaustive search at the five power levels. The existing analytical model provides an approximation error of 25%. Next, our proposed framework highlights that the optimal per-level offered load restrictively improves the throughput above the optimally uniform per-level offered load. Finally, our proposed framework discovers a 27% more energy-efficient per-level offered load than the existing framework at the five power levels while providing higher throughput than the optimally uniform per-level offered load.