Optimal Task Offloading and Resource Allocation in Software-Defined Vehicular Edge Computing

In vehicular edge computing (VEC), resource-intensive tasks are offloaded to computing nodes at the network edge. Owing to high mobility and distributed nature, optimal task offloading in vehicular environments is still a challenging problem. In this paper, we first introduce a software-defined vehicular edge computing (SD-VEC) architecture where a controller not only guides the vehicles' task offloading strategy but also determines the edge cloud resource allocation strategy. To obtain the optimal strategies, we formulate a problem on the edge cloud selection and resource allocation to maximize the probability that a task is successfully completed within a pre-specified time limit. Since the formulated problem is a well-known NP-hard problem, we devise a mobility-aware greedy algorithm (MGA) that determines the amount of edge cloud resources allocated to each vehicle. Trace-driven simulation results demonstrate that MGA provides near-optimal performance and improves the successful task execution probability compared with conventional algorithms.