Sub-Band Full Duplex Resource Allocation: A Predictive Deep Reinforcement Learning Approach

2026-05-14

AI Summary

This work addresses the challenge of balancing uplink and downlink performance in subband full-duplex systems under dynamic traffic conditions by proposing a proactive resource scheduling framework that integrates high-accuracy traffic forecasting with deep reinforcement learning. Specifically, a hybrid bidirectional LSTM model predicts future traffic loads, and a Double Deep Q-Network (Double DQN) leverages these predictions together with current queue states to dynamically optimize subband allocation between uplink and downlink. By uniquely combining forward-looking traffic prediction with real-time decision-making, the proposed approach overcomes the limitations of conventional static or reactive resource management schemes. Experimental results demonstrate that this method significantly enhances spectral efficiency and effectively reduces queue backlogs, exhibiting superior adaptability and performance for autonomous resource management in 6G networks.

Abstract

This paper presents a predictive deep learning framework for dynamic sub-band allocation in Sub-Band Full Duplex (SBFD) systems, addressing the challenge of balancing uplink (UL) and downlink (DL) performance under highly dynamic traffic conditions. The key contribution lies in integrating a hybrid Bidirectional Long Short-Term Memory (Bi-LSTM) model for traffic forecasting with a Double Deep Q-Network (DDQN) for real-time resource allocation. Using both predicted traffic and current queue states, the proposed system enables proactive scheduling based on traffic demand. Evaluation results show that the prediction model achieves high accuracy in capturing bursty traffic patterns, while the DDQN agent effectively adapts UL/DL split ratios according to traffic variations. The framework improves spectrum utilization, reduces queue buildup, and avoids inefficient static configurations. The proposed approach demonstrates that combining predictive intelligence with reinforcement learning significantly enhances the efficiency and adaptability of SBFD systems, making it a strong candidate for autonomous resource management in future 6G networks.

Problem

• Sub-Band Full Duplex
• resource allocation
• dynamic traffic
• uplink-downlink balancing
• spectrum utilization

Innovation

• Predictive Deep Reinforcement Learning
• Sub-Band Full Duplex
• Bi-LSTM
• Double Deep Q-Network
• Dynamic Resource Allocation