Joint Activity Detection and Channel Estimation for 6G GFRA: A Memory-Enhanced DL Network Framework
Zhen, Li Fan, Yuanbo Cheng, Luyao Li, Mingyang Li, Shuchang Lu, Guangyue Guizani, Mohsen
Zhen, Li
Fan, Yuanbo
Cheng, Luyao
Li, Mingyang
Li, Shuchang
Lu, Guangyue
Guizani, Mohsen
Supervisor
Department
Machine Learning
Embargo End Date
Type
Workshop
Date
2025
License
Language
English
Collections
Research Projects
Organizational Units
Journal Issue
Abstract
Grant-free random access (GFRA), owning to its short access delay and low signaling overhead, has been regarded as a promising enabling technique to support massive access in 6G Internet of Things (IoT). The key of GFRA lies in achieving accurate joint activity detection and channel estimation (JADCE), which can be viewed as a typical sparse signal recovery (SSRe) problem due to the sporadic traffic of IoT devices. To solve this problem, the conventional compressed sensing algorithms can be utilized, which, however, suffer from considerable computational complexities and lack robustness to diverse sensing matrices. In this paper, we propose a memory-augmented iterative shrinkage thresholding algorithm network framework for JADCE, which efficiently enhances SSRe performance by learning sprasifying transforms and reinforcing information interaction in the residual domain. Particularly, we introduce a high-throughput short-term memory mechanism into the unfolded multi-stage deep network, so as to guarantee both the information abundance and interpretability of the network. Simulation results indicate the significant superiority of proposed network in terms of the detection and estimation performance over the existing model-driven deep unfolding networks.
Citation
L. Zhen et al., "Joint Activity Detection and Channel Estimation for 6G GFRA: A Memory-Enhanced DL Network Framework," 2024 IEEE Globecom Workshops (GC Wkshps), Cape Town, South Africa, 2024, pp. 1-6, doi: 10.1109/GCWkshp64532.2024.11101207
Source
2024 IEEE Globecom Workshops (GC Wkshps)
Conference
IEEE Globecom Workshops (GC Wkshps)
Keywords
6G, Grant-Free Random Access, Joint Activity Detection and Channel Estimation, Iterative Shrinkage Thresholding Algorithm, High-Throughput Short-Term Memory
Subjects
Source
IEEE Globecom Workshops (GC Wkshps)
Publisher
IEEE