Joint Computation Offloading and Resource Allocation for Hybrid Cloud and Edge Computing in Satellite-assisted Unmanned System
Yeyu Wu, Huili Fan, Jian Yu, Yang Li, Qilong Huang, Yaowen Qi · 2024
Unmanned systems such as unmanned surface vehicle (USV) and unmanned aerial vehicle (UAV) are difficult to undertake complex computational tasks on their own due to their limited computational capabilities. The existing method of sending the computation tasks to the ground terminal for processing will bring large transmission delay and reduce the real-time performance of task execution. Inspired by terrestrial mobile edge computing (MEC), deploying edge computing servers on low-orbit satellites (LEO) allows unmanned system to offload tasks to the satellite to enhance the task processing efficiency. However, owing to restricted computing and communication resources of satellite, designing appropriate computational offloading and resource allocation strategies is of great relevance. In this article, we propose a partial computation offloading strategy for a three-tier edge computing architecture consisting of unmanned system, satellites and cloud. We consider the energy optimization of the system under resource constraints while meeting the requirements of tasks. To address this issue, we propose a computational offloading method based on Deep Deterministic Policy Gradient (DDPG). This method can simultaneously decide the optimal offloading ratio, offloading location and resource allocation ratio under the dynamic environment with multi-task concurrency. The proposed method meets task’s requirements while effectively decreasing the total energy consumption of system. Finally, experimental results prove the effectiveness of the formulated method.