Digital Twin-Assisted Task Offloading and Resource Allocation in UAV-Enabled Mobile Vehicular Edge Computing With Energy Harvesting

This paper investigates a digital twin-assisted Uncrewed Aerial Vehicle (UAV)-enabled Mobile Vehicular Edge Computing (UMVEC) network, where energy harvesting is introduced to address the critical issue of UAV energy scarcity. Wireless power transfer is integrated to enable sustainable UAV operations. We aim to minimize the long-term weighted sum of task processing delay and energy consumption for multiple tasks from vehicles, subject to constraints on computing capabilities, energy capabilities, and task requirements. To achieve this, we formulate a joint optimization problem involving task offloading, computing resource allocation, and UAV charging strategies. The computing resource allocation strategy from RSU and UAV, the task offloading strategy from vehicles, and the charging strategy for UAV are determined together in digital twin-assisted UMVEC. To solve this complex problem, a multi-agent deep reinforcement learning algorithm based on the centralized training with decentralized execution paradigm is proposed. Furthermore, to enhance learning efficiency, we incorporate a variance-decaying exploration mechanism that adaptively balances exploration and exploitation throughout the training process. Simulation results demonstrate that the proposed digital twin-assisted scheme significantly outperforms three baselines, achieving average performance gains of 14.57%, 28.77%, and 30.2%, respectively.