Task-Offload-Oriented Resource Optimization in MEC-Enabled System with Multiple BSs

In the extensive research on next-generation communication network architectures, task offloading and resource allocation problems become increasingly complex for mobile edge computing (MEC) systems with multiple base stations (BSs). This paper first formulates this multi-dimensional dynamic issue as an optimization problem. To minimize system overhead, we propose an attention-based multi-agent proximal policy optimization (A-MAPPO) algorithm. This algorithm employs a centralized training and decentralized execution (CTDE) framework, and leverages attention mechanisms to facilitate the convergence of the critic network, thereby enhancing the algorithm's performance. Experimental results show that the A-MAPPO algorithm can reduce system costs by up to 28.3% compared to other benchmark algorithms.