A MARL-Based Approach for Massive Access in IRS-Aided NOMA-URLLC Networks
Han, Huimei Wang, Hongyang Lu, Weidang Zhai, Wenchao Li, Ying Wu, Celimuge Guizani, Mohsen
Han, Huimei
Wang, Hongyang
Lu, Weidang
Zhai, Wenchao
Li, Ying
Wu, Celimuge
Guizani, Mohsen
Supervisor
Department
Machine Learning
Embargo End Date
Type
Journal article
Date
2025
License
Language
English
Collections
Research Projects
Organizational Units
Journal Issue
Abstract
Ultra-reliable low-latency communication (URLLC) supports services that require stringent low latency and high reliability, as well as finite block transmission. The goal of the future sixth-generation (6G) networks is to implement the internet-of-everything, where the number of URLLC users is expected to reach the order of millions. Intelligent reflective surfaces (IRS) and non-orthogonal multiple access (NOMA) technologies can enhance the performance of URLLC communications by adjusting wireless channels and allowing multiple users in the same resource block, respectively. In this paper, to manage massive users in IRS-aided NOMA URLLC networks, the resource assignment strategy (including sub-channel allocation, transmitting power selection, and the phases of IRS units) is optimized using a proposed multi-agent reinforcement learning (MARL)-based algorithm, while meeting the reliability and latency demands of URLLC services. In addition, transfer learning is introduced to reduce learning overheads and enhance the probability of successful access. Our simulation results indicate that the proposed MARL-based approach significantly outperforms baseline methods in terms of the successful access probability for scenarios with massive users.
Citation
H. Han et al., "A MARL-Based Approach for Massive Access in IRS-Aided NOMA-URLLC Networks," in IEEE Transactions on Cognitive Communications and Networking, doi: 10.1109/TCCN.2025.3643938
Source
IEEE Transactions on Cognitive Communications and Networking
Conference
Keywords
Massive access, multi-agent reinforcement learning, URLLC, wireless communication
Subjects
Source
Publisher
IEEE