Multiuser Activity and Data Detection via Sparsity-Blind Greedy Recovery for Uplink Grant-Free NOMA

Abstract

Exploiting the sparse activity of users, compressed sensing (CS) has been of interest in multiuser detection (MUD) for non-orthogonal multiple access (NOMA), to enable a massive connection of users in machine-type communications (MTC). In this letter, we propose a greedy algorithm with unknown sparsity level for CS-based multiuser activity and data detection in uplink grant-free NOMA. To accommodate practical scenarios, the algorithm employs a criterion to stop the iteration with no prior knowledge of sparsity level. Also, it requires no knowledge of noise variance by computing the log-likelihood ratios (LLR) approximately in its operation. With no need of sparsity and noise levels, we perform CS-based MUD with low complexity. Simulation results demonstrate that the proposed algorithm outperforms conventional CS-based solutions, nearly achieving the oracle performance of fully known supports.