Cognitively-Inspired Two-Stage Diffusion Policy for Adaptive Resource Scheduling in Sustainable UAV-Assisted Mobile Edge Computing Systems
TL;DR
A TS-Diff (Two-Stage Diffusion Policy) framework for joint task offloading, trajectory planning, and energy harvesting, integrating diffusion-based policy generation with reinforcement learning, provides a cognitively inspired decision-making approach for intelligent UAV-MEC systems in dynamic environments.
Abstract
In Unmanned Aerial Vehicle-assisted Mobile Edge Computing (UAV-MEC), dynamic workloads and limited onboard energy pose significant challenges for efficient task scheduling and long-term mission sustainability. Cognitively-inspired computing paradigms provide an intelligent solution by enabling UAVs to perceive environments, learn from experience, and make adaptive decisions. This paper proposes a TS-Diff (Two-Stage Diffusion Policy) framework for joint task offloading, trajectory planning, and energy harvesting. A brief Soft Actor-Critic pre-training stage first constructs an exploratory experience memory buffer to address the cold-start issue of diffusion models. A Diffusion Policy Actor is then employed to iteratively generate robust continuous control actions, forming a perception–decision–action loop for adaptive UAV control. Experimental results show that TS-Diff achieves a final average return of approximately -145, improving performance by about 20% compared with baseline algorithms. The framework also increases total task throughput to 971.6, significantly outperforming DDPG (770.3), while enabling adaptive charging strategies that prevent energy depletion. By integrating diffusion-based policy generation with reinforcement learning, the proposed framework provides a cognitively inspired decision-making approach for intelligent UAV-MEC systems in dynamic environments.