Ultra-Reliable Computation Offloading for Aerial-Ground Cooperative Vehicular Networks with Joint Mobility Optimization and Transmission Scheduling

The reliability of computation offloading poses a challenge to aerial ground cooperative vehicular networks (AGCVNs) due to the highly dynamic and stochastic nature of ground-to-air (G2A) channels and channel interferences. In this paper, we develop a joint optimization model to maximize the reliability of computation offloading in unmanned aerial vehicle (UAV)-assisted and mobile edge computing (MEC)-enabled AGCVNs. Specifically, we jointly optimize the mobility of a UAV-mounted cloudlet and the data transmission scheduling of a ground vehicle that would like to offload computation tasks to the UAV under a data integrity constraint and a time horizon constraint. We theoretically characterize the distribution of the signal to interference plus noise ratio (SINR) of the G2A channel regarding stochastic interferences and derive the expected offloading capacity and an upper bound of the variance of the offloaded data volume in the channel. We also derive a lower bound of the offloading reliability that the vehicle can successfully offload all the task data to the UAV via the G2A channel within a given time period. The lower bound can provide a mathematically tractable formulation of the reliability optimization objective for the network, which leads to a joint mobility optimization and data transmission scheduling method. We validate the derived theoretical models via simulations. The simulation results also show that the proposed joint optimization method achieves the offloading reliability of 99.4826% on average, outperforming the conventional computation offloading scheme by enabling the adaptive response of computation offloading to the network mobility.