Dynamic Resource Allocation in Cell-Free MEC: Leveraging Attention Mechanism with MADDPG

This paper presents a novel approach for dynamic resource allocation in cell-free massive MIMO enabled mobile edge computing (MEC) systems, employing an attention mechanism within a Multi-Agent Deep Deterministic Policy Gradient (MADDPG) framework to adaptively weigh the contributions of other agents. This approach allows for a more nuanced understanding of the interactions between agents, leading to a more efficient resource distribution. We demonstrate the effectiveness of the proposed algorithm in meeting delay requirements while reducing energy consumption of the users. Notably, our scheme exhibits robust generalization capabilities, effectively accommodating varying number of users beyond the training scenario.