Graph Attention Network-Based Multi-Agent Reinforcement Learning for Slicing Resource Management in Dense Cellular Network

Network slicing (NS) management devotes to providing various services meet distinct requirements over the same physical communication infrastructure and allocating resources on demands. Considering a dense cellular network scenario that contains several NS multiple base stations (BSs), it remains challenging design proper real-time inter-slice resource strategy, so as cope with frequent BS handover satisfy fluctuations of service requirements. In this paper, we propose formulate challenge multi-agent reinforcement learning (MARL) problem in which each represents an agent. Then, leverage graph attention (GAT) strengthen temporal spatial cooperation between agents. Furthermore, incorporate GAT into deep (DRL) correspondingly intelligent strategy. More specially, testify universal effectiveness for advancing DRL system, by applying top both value-based method Q-network (DQN) combination policy-based advantage actor-critic (A2C). Finally, verify superiority GAT-based MARL algorithms through extensive simulations.