Semi-Grant-Free NOMA: A Stochastic Geometry Model

Chao Zhang, Yuanwei Liu, Zhiguo Ding

Research output: Contribution to journalArticlepeer-review

17 Scopus citations

Abstract

Grant-free (GF) transmission holds promise in terms of low latency communication by directly transmitting messages without waiting for any permissions. However, collision situations may frequently happen when limited spectrum is occupied by numerous GF users. The non-orthogonal multiple access (NOMA) technique can be a promising solution to achieve massive connectivity and fewer collisions for GF transmission by multiplexing users in power domain. We utilize a semi-grant-free (semi-GF) NOMA scheme for enhancing network connectivity and spectral efficiency by enabling grant-based (GB) and GF users to share the same spectrum resources. With the aid of semi-GF protocols, uplink NOMA networks are investigated by invoking stochastic geometry techniques. We propose a novel dynamic protocol to interpret which part of the GF users are allocated in NOMA transmissions via transmitting various channel quality thresholds by an added handshake. We utilize open-loop protocol with a fixed average threshold as the benchmark to investigate performance improvement. It is observed that dynamic protocol provides more accurate channel quality thresholds than open-loop protocol, thereby the interference from the GF users is reduced to a large extent. We analyze the outage performance and diversity gains under two protocols. Numerical results demonstrate that dynamic protocol is capable of enhancing the outage performance than open-loop protocol.

Original languageBritish English
Pages (from-to)1197-1213
Number of pages17
JournalIEEE Transactions on Wireless Communications
Volume21
Issue number2
DOIs
StatePublished - 1 Feb 2022

Keywords

  • Dynamic protocol
  • grant-based
  • grant-free
  • stochastic geometry
  • uplink NOMA

Fingerprint

Dive into the research topics of 'Semi-Grant-Free NOMA: A Stochastic Geometry Model'. Together they form a unique fingerprint.

Cite this