Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Reinforcement Learning-Based Dynamic Resource Allocation in Cell-Free Massive MIMO

Phuong Nam Tran, Nhan Thanh Nguyen, Hien Quoc Ngo, Markku Juntti

TL;DR

This work tackles EE optimization in CFmMIMO by deriving closed-form SE and EE expressions and proposing a scalable DRL framework that maps large-scale fading to a fixed-size action space of AP activation, antenna allocation, and power control. By computing antenna counts and power factors from simple closed-form rules, the method achieves efficient, real-time capable resource allocation. The PPO-based DRL agent attains around 50% EE gains and orders-of-magnitude runtime reductions compared with conventional SCA methods, highlighting substantial practical impact for large-scale deployments. Overall, the approach enables energy-efficient CFmMIMO operation through a scalable, learning-based, cross-layer optimization strategy.

Abstract

In this paper, we consider power allocation and antenna activation of cell-free massive multiple-input multiple-output (CFmMIMO) systems. We first derive closed-form expressions for the system spectral efficiency (SE) and energy efficiency (EE) as functions of the power allocation coefficients and the number of active antennas at the access points (APs). Then, we aim to enhance the EE through jointly optimizing antenna activation and power control. This task leads to a non-convex and mixed-integer design problem with high-dimensional design variables. To address this, we propose a novel DRL-based framework, in which the agent learns to map large-scale fading coefficients to AP activation ratio, antenna coefficient, and power coefficient. These coefficients are then employed to determine the number of active antennas per AP and the power factors assigned to users based on closed-form expressions. By optimizing these parameters instead of directly controlling antenna selection and power allocation, the proposed method transforms the intractable optimization into a low-dimensional learning task. Our extensive simulations demonstrate the efficiency and scalability of the proposed scheme. Specifically, in a CFmMIMO system with 40 APs and 20 users, it achieves a 50% EE improvement and 3350 times run time reduction compared to the conventional sequential convex approximation method.

Deep Reinforcement Learning-Based Dynamic Resource Allocation in Cell-Free Massive MIMO

TL;DR

This work tackles EE optimization in CFmMIMO by deriving closed-form SE and EE expressions and proposing a scalable DRL framework that maps large-scale fading to a fixed-size action space of AP activation, antenna allocation, and power control. By computing antenna counts and power factors from simple closed-form rules, the method achieves efficient, real-time capable resource allocation. The PPO-based DRL agent attains around 50% EE gains and orders-of-magnitude runtime reductions compared with conventional SCA methods, highlighting substantial practical impact for large-scale deployments. Overall, the approach enables energy-efficient CFmMIMO operation through a scalable, learning-based, cross-layer optimization strategy.

Abstract

In this paper, we consider power allocation and antenna activation of cell-free massive multiple-input multiple-output (CFmMIMO) systems. We first derive closed-form expressions for the system spectral efficiency (SE) and energy efficiency (EE) as functions of the power allocation coefficients and the number of active antennas at the access points (APs). Then, we aim to enhance the EE through jointly optimizing antenna activation and power control. This task leads to a non-convex and mixed-integer design problem with high-dimensional design variables. To address this, we propose a novel DRL-based framework, in which the agent learns to map large-scale fading coefficients to AP activation ratio, antenna coefficient, and power coefficient. These coefficients are then employed to determine the number of active antennas per AP and the power factors assigned to users based on closed-form expressions. By optimizing these parameters instead of directly controlling antenna selection and power allocation, the proposed method transforms the intractable optimization into a low-dimensional learning task. Our extensive simulations demonstrate the efficiency and scalability of the proposed scheme. Specifically, in a CFmMIMO system with 40 APs and 20 users, it achieves a 50% EE improvement and 3350 times run time reduction compared to the conventional sequential convex approximation method.
Paper Structure (15 sections, 19 equations, 2 figures, 1 algorithm)

This paper contains 15 sections, 19 equations, 2 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. System Model
  3. Channel Estimation
  4. Downlink Data Transmission
  5. Problem Formulation
  6. Spectral Efficiency
  7. Power Consumption
  8. EE Maximization Problem
  9. DRL for Resource Allocation
  10. Resource Allocation
  11. DRL Agent Formulation
  12. Proximal Policy Optimization
  13. PPO-based Algorithm for Resource Allocation
  14. Performance Evaluation
  15. Conclusion

Figures (2)

  • Figure 1: Cell-free massive MIMO system.
  • Figure 2: Convergence, energy efficiency, and runtime performance of the proposed DRL scheme compared with benchmarks.