Unsourced Random Access with a Massive MIMO Receiver: Exploiting Angular Domain Sparsity

Abstract

This paper investigates the unsourced random access (URA) scheme to accommodate a large amount of machine-type users communicating to a massive MIMO base station. Existing works adopt a slotted transmission strategy to reduce system complexity and operate under the framework of coupled compressed sensing (CCS), concatenating an outer tree code to an inner compressed sensing code for message stitching. We observe that the sparse angular domain MIMO channel can help decouple the CCS scheme and introduce an uncoupled slotted transmission scheme without the tree encoder/decoder. We propose a novel MRF-GAMP method capturing the structured sparsity of the angular domain channel for activity detection and channel estimation. Then, message reconstruction is based on rearranging strongly correlated slot-wise channels into groups by a clustering algorithm. Extensive simulation shows that our approach achieves a better error performance and a higher spectral efficiency compared to the CCS scheme.