Proximal Policy Optimization for Up/Downlink Time Slots Allocation in 5/6G Dynamic TDD Networks

• Trang Nguyen T. H.,
• Taejoon Kim,

Vol. 19, No. 1, January 31, 2025
10.3837/tiis.2025.01.012, Download Paper (Free):

• TDD configuration
• policy gradients
• Proximal Policy Optimization (PPO)
• Resource Allocation
• Deep Reinforcement Learning

Abstract

In recent years, the significant development of 5G and beyond networks has led to a notable increase in user demand for up/downlink data rates in mobile communication systems, giving rise to challenges due to the limited availability of radio resources. To address this issue, an effective up/downlink time slots allocation has been investigated in dynamic time division duplex (TDD) based mobile wireless networks. We propose a novel method leveraging the proximal policy optimization (PPO) algorithm, a state-of-the-art technique in deep reinforcement learning, to intelligently schedule up/downlink TDD time slots. The proposed algorithm dynamically assigns up/downlink time slots in response to the changing numbers of up/downlink requests, diverse quality of service (QoS) requirements, and mobility scenarios, aiming to maximize both uplink and downlink sum-rates while ensuring minimum up/downlink data rates are satisfied. Simulation results demonstrate that the proposed algorithm achieves a superior network performance in terms of the data sum-rate compared to other benchmark approaches in dynamic TDD networks.

Statistics

Cite this article