Fine-Grained Task Offloading for UAV-Assisted Mobile Edge Computing

To solve the challenge of task offloading decisions and adaptive link selection of air-to-ground channels for UAV-assisted mobile edge computing, a fine-grained task offloading method is proposed. Firstly, an air-to-ground communication channel model is designed considering the probability of line-of-sight (LOS) and non-line-of-sight (NLOS) links in different signal propagation environments. On this basis, the task offloading optimization objective is defined to minimize the weighted value of delay energy consumption, and the impact of different weighted parameters on the task offloading algorithm is considered. Finally, a dynamic task offloading approach based on the depth deterministic strategy gradient algorithm is presented, and an efficient offloading scheme is obtained through adaptive learning of dynamic air-to-ground networks. Simulation results demonstrate that the proposed approach effectively mitigates system energy consumption and reduces computational latency within UAV-assisted mobile edge computing systems, ensure system efficiency, stability and reliability, and better meet the service requirements of mobile edge users compared to other benchmark algorithms.