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 GraphConvolutional Network (GCN) integrated with a Deep Q-Network (DQN)reinforcement learning framework for improved interference coordination inmulti-cell, multi-channel networks. Unlike LDP's static priorities, ourapproach dynamically learns link priorities based on real-time traffic demand,network topology, remaining transmission opportunities, and interferencepatterns. The GCN captures spatial dependencies, while the DQN enables adaptivescheduling decisions through reward-guided exploration. Simulation results showthat our GCN-DQN model achieves mean SINR improvements of 179.6\%, 197.4\%, and175.2\% over LDP across three network configurations. Additionally, the GCN-DQNmodel demonstrates mean SINR improvements of 31.5\%, 53.0\%, and 84.7\% overour previous CNN-based approach across the same configurations. These resultsunderscore the effectiveness of our GCN-DQN model in addressing complex URLLCrequirements with minimal overhead and superior network performance.