From Simulation to Reality: Practical Deep Reinforcement Learning-based Link Adaptation for Cellular Networks

2026-02-28

AI Summary

This work addresses the performance degradation of existing deep reinforcement learning (DRL)-based link adaptation methods when deployed in real-world systems, as they are typically evaluated under idealized simulation conditions that neglect practical impairments such as ACK/NACK feedback delays, HARQ retransmissions, and DRL inference latency. To bridge this gap, the authors propose the Decoupling-DQN framework (DC-DQN-LA), which uniquely decouples DRL training and inference into a non-real-time training module and a real-time inference module. The design explicitly incorporates feedback delay and parallel HARQ mechanisms into the state, action, and reward formulations. Validated on the srsRAN 5G protocol stack with a USRP software-defined radio platform, the approach achieves 40%–70% higher throughput than baseline algorithms under mobility scenarios while maintaining comparable block error rates and demonstrating rapid adaptation to dynamic channel conditions, thereby significantly enhancing real-world deployment feasibility.

Technology Category

• Machine Learning: Hardware-aware ML
• Search and Optimization: Learning to Search
• Natural Language Processing: Learning & Optimization for NLP

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Link Adaptation (LA) that dynamically adjusts the Modulation and Coding Schemes (MCS) to accommodate time-varying channels is crucial and challenging in cellular networks. Deep reinforcement learning (DRL)-based LA that learns to make decision through the interaction with the environment is a promising approach to improve throughput. However, existing DRL-based LA algorithms are typically evaluated in simplified simulation environments, neglecting practical issues such as ACK/NACK feedback delay, retransmission and parallel hybrid automatic repeat request (HARQ). Moreover, these algorithms overlook the impact of DRL execution latency, which can significantly degrade system performance. To address these challenges, we propose Decoupling-DQN (DC-DQN), a new DRL framework that separates traditional DRL's coupled training and inference processes into two modules based on Deep Q Networks (DQN): a real-time inference module and an out-of-decision-loop training module. Based on this framework, we introduce a novel DRL-based LA algorithm, DC-DQN-LA. The algorithm incorporates practical considerations by designing state, action, and reward functions that account for feedback delays, parallel HARQ, and retransmissions. We implemented a prototype using USRP software-defined radios and srsRAN software. Experimental results demonstrate that DC-DQN-LA improves throughput by 40\% to 70\% in mobile scenario compared with baseline LA algorithms, while maintaining comparable block error rates, and can quickly adapt to environment changes in mobile-to-static scenario. These results highlight the efficiency and practicality of the proposed DRL-based LA algorithm.

Problem

• Link Adaptation
• Deep Reinforcement Learning
• Feedback Delay
• HARQ
• Execution Latency

Innovation

• Decoupling-DQN
• Link Adaptation
• Deep Reinforcement Learning
• HARQ
• Practical Wireless Implementation