AoI minimization for Uplink Cell-Free Networks: A DRL-Based Multi-Objective Approach

TY - GEN

T1 - AoI minimization for Uplink Cell-Free Networks

T2 - 2024 IEEE Middle East Conference on Communications and Networking, MECOM 2024

AU - Waraiet, Abdulhamed

AU - Cumanan, Kanapathippillai

AU - Rehan, Salahedin

AU - Al-Shami, Tareq

AU - Grace, David

AU - Burr, Alister

N1 - Publisher Copyright: © 2024 IEEE.

PY - 2025/2/18

Y1 - 2025/2/18

N2 - In this paper, we propose a deep reinforcement learning (DRL)-based framework to jointly minimize the ergodic age of information (AoI) and the total transmit power in an uplink (UL) cell-free massive multiple-input multiple-output (CF-mMIMO) system. In particular, the multiple-objective resource allocation problem is formulated into an optimization problem subject to quality-of-service (QoS) and maximum transmit power constraints. Due to the long-term nature of the problem, it is challenging to solve using conventional convex optimization techniques. Therefore, the problem is reformulated as a reinforcement learning (RL) environment and a novel state space and reward function are developed. Finally, the soft actor-critic DRL agent is developed to solve the reformulated problem. Simulation results demonstrate that the proposed scheme achieves significant power savings while maintaining a relatively low average AoI score compared to the benchmark schemes.

AB - In this paper, we propose a deep reinforcement learning (DRL)-based framework to jointly minimize the ergodic age of information (AoI) and the total transmit power in an uplink (UL) cell-free massive multiple-input multiple-output (CF-mMIMO) system. In particular, the multiple-objective resource allocation problem is formulated into an optimization problem subject to quality-of-service (QoS) and maximum transmit power constraints. Due to the long-term nature of the problem, it is challenging to solve using conventional convex optimization techniques. Therefore, the problem is reformulated as a reinforcement learning (RL) environment and a novel state space and reward function are developed. Finally, the soft actor-critic DRL agent is developed to solve the reformulated problem. Simulation results demonstrate that the proposed scheme achieves significant power savings while maintaining a relatively low average AoI score compared to the benchmark schemes.

KW - age of information

KW - CF-mMIMO

KW - DRL

KW - power control

KW - SAC

UR - http://www.scopus.com/inward/record.url?scp=86000263451&partnerID=8YFLogxK

U2 - 10.1109/MECOM61498.2024.10880858

DO - 10.1109/MECOM61498.2024.10880858

M3 - Conference contribution

AN - SCOPUS:86000263451

T3 - 2024 IEEE Middle East Conference on Communications and Networking, MECOM 2024

SP - 315

EP - 320

BT - 2024 IEEE Middle East Conference on Communications and Networking, MECOM 2024

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 17 November 2024 through 20 November 2024

ER -