Deep Learning-Based Task Offloading for Vehicular Edge Computing.

With widely used of computation intensive vehicular applications, Vehicular Edge Computing (VEC) plays an increasing important role in providing computation service for vehicles, and task scheduling has direct impact on task offloading performance. In this paper, we analyze the scheduling of tasks in VEC, propose a deep learning-based task scheduling mechanism, and design a model suitable for the prediction of task scheduling success probability and offloading delay. The performance of the model is evaluated with a large amount of data compared to the SVM-based task scheduling algorithm, the proposed mechanism has the offloading failure rate decreased by 40%.