Temporally Consistent Graph Q-Networks for Intelligent Network Control
Researchers have developed a novel multi-agent reinforcement learning algorithm called the Temporally Consistent Graph Q-Network (TC-GQN) for optimizing mobile network control. This algorithm learns a task-independent representation of the entire network, aggregating information from all base stations. A graph neural network then uses this encoding to coordinate local actions based on a global reward function, demonstrating improved hardware sleep time while maintaining quality of service compared to existing baselines. AI