Edge-Based V2X communications with big data intelligence

Vehicular Internet-of-Things applications require an efficient Vehicle-to-Everything (V2X) communication scheme. However, it is particularly challenging to achieve a high throughput and low latency with limited wireless resources in highly dynamic vehicular networks. In this article, we propose a scheme that enhances V2V communications through integration of vehicle edge-based forwarding and learning-based edge selection policy optimization. The proposed scheme has three main characteristics. First, the Hierarchical edge-based preemptive route creation is introduced to create hierarchical edges and conduct efficient packet forwarding as well as route aggregation. Second, Two-stage learning is introduced to select efficient edge nodes using big data driven traffic prediction and reinforcement learning-based edge node selection. Third, Context-aware edge selection is employed to improve the performance of edge-based forwarding in various contexts. We use real traffic big data and realistic vehicular network simulations to evaluate the performance of the proposed scheme and show the advantage over other baseline approaches.