Determinantal Point Process-Based Learning for GF User Selection in Semi-Grant-Free NOMA Systems

Abstract

In this letter, based on the Determinantal Point Process (DPP)-based learning (DPPL), we provide an efficient solution for the multiple grant-free (GF) users scheduling problem in the Non-orthogonal multiple access (NOMA) assisted semi-grant-free (SGF) transmission scheme with hybrid successive interference cancellation (HSIC). This solution selects appropriate GF users to share the grant-based (GB) user’s channel by considering both the diversity measure and the quality measure, so as to optimize the sum of the throughput of selected GF users. Numerical results demonstrate that the proposed GF user scheduling scheme based on DPPL can provide a higher sum of admitted GF users’ throughput than the existing GF user scheduling schemes in the SGF, and its running time is much shorter than that of the popular optimization algorithm.