Online Distributed Optimization for Energy-Efficient Computation Offloading in Air-Ground Integrated Networks

Driven by ever-increasing vehicular intelligent computation-intensive and delay-sensitive services, this paper in-vestigates the computing offloading in unmanned aerial vehicle (UAV)-assisted vehicular networks. Due to the limited onboard energy and computational resources of the mobile entities (i.e., the vehicles and the UAV), it is significant to explore the collab-orative computation among the vehicles, the UAV, and the ter-restrial computing servers for improving energy efficiency (EE) while trading off the service delay. Unlike existing work in the literature that is based on offline settings with a global view, an online distributed mechanism is proposed to cope with the spatial and temporal variations of the networks. Specifically, upon the arriving tasks and the real-time channel conditions, mobile entities adaptively decide about the task offloading and computational resources allocation in parallel. Moreover, the UAV also designs its trajectory with the residual battery capacity taken into account. Theoretical analysis shows that the developed approach can achieve the EE-delay tradeoff as [O(1/V ), O (V )] with V being a control parameter, and can strike a flexible balance between them by tuning V. Numerical results verify the theoretical analysis and reveal that the performance gain can be obtained over conventional methods in the EE performance.