MADDPG-Based Joint Latency and Energy Optimization for UE Aggregation Cooperative Transmission

The surging demand for high-reliability and lowlatency services in the Internet of Everything (IoE), smart cities, etc., challenges the network resource optimization under dense terminal access. To address this problem, this paper proposes a dynamic clustering algorithm to support user equipments (UEs) aggregation, where cooperative nodes (CNs) help to transfer data from UEs with poor coverage or limited capabilities (e.g., low-power devices) to the base station. The proposed dynamic clustering algorithm adapts UE aggregation to environmental characteristics for different scenario optimization. Furthermore, a Multi-Agent Deep Deterministic Policy Gradient (MADDPG) strategy is designed to achieve joint latency-energy optimization, since there are multiple CNs within a single coverage. To validate the effectiveness of the dynamic clustering and MADDPGbased CN optimization, ablation experiments are designed to verify the performance improvements of the proposed scheme. Simulation results show that this scheme effectively achieves joint optimization of latency and energy consumption in high-density UE communication environment.