GCN-Driven Reinforcement Learning for Probabilistic Real-Time Guarantees in Industrial URLLC

2025-06-17

AI Summary

To address the challenge of achieving packet-level ultra-reliable low-latency communication (URLLC) reliability and latency guarantees in large-scale industrial multi-cell, multi-channel wireless networks, this paper proposes a joint GCN-DQN scheduling framework. It employs Graph Convolutional Networks (GCNs) to model topological structures and interference coupling among base stations and users, while integrating Deep Q-Networks (DQNs) for real-time link prioritization based on instantaneous traffic, residual resources, and SINR-aware channel conditions. This work is the first to synergistically combine GCNs and DQNs for URLLC interference coordination, overcoming the limitations of conventional static, locally deadline-driven priority assignment (LDP). The framework enables spatiotemporally coupled dynamic optimization. Experimental results demonstrate that, compared to the LDP baseline, the proposed method improves average SINR by 175.2%–197.4%; against a CNN-based baseline, it achieves gains of 31.5%–84.7%, significantly enhancing URLLC performance.

Technology Category

• Planning, Routing, and Scheduling: Scheduling under Uncertainty
• Machine Learning: Graph-based Machine Learning
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Ensuring packet-level communication quality is vital for ultra-reliable, low-latency communications (URLLC) in large-scale industrial wireless networks. We enhance the Local Deadline Partition (LDP) algorithm by introducing a Graph Convolutional Network (GCN) integrated with a Deep Q-Network (DQN) reinforcement learning framework for improved interference coordination in multi-cell, multi-channel networks. Unlike LDP's static priorities, our approach dynamically learns link priorities based on real-time traffic demand, network topology, remaining transmission opportunities, and interference patterns. The GCN captures spatial dependencies, while the DQN enables adaptive scheduling decisions through reward-guided exploration. Simulation results show that our GCN-DQN model achieves mean SINR improvements of 179.6%, 197.4%, and 175.2% over LDP across three network configurations. Additionally, the GCN-DQN model demonstrates mean SINR improvements of 31.5%, 53.0%, and 84.7% over our previous CNN-based approach across the same configurations. These results underscore the effectiveness of our GCN-DQN model in addressing complex URLLC requirements with minimal overhead and superior network performance.

Problem

• Enhancing URLLC packet-level quality in industrial wireless networks
• Dynamic link priority learning for multi-cell interference coordination
• Improving SINR performance via GCN-DQN adaptive scheduling

Innovation

• GCN-DQN model for dynamic link prioritization
• Real-time traffic and interference adaptation
• Spatial dependency capture with GCN