5G-Activated VANETs: Enormous Edge Intelligence for AI Integrated Dynamic Routing using Ultra-Low Latency Vehicle-to-Everything (V2X) Communication

Abstract

With the rapid deployment of autonomous and connected vehicles, stringent requirements have been placed on ultra-reliable low latency communication systems for safe, reliable and intelligent transportation systems. The fifth-generation (5G) wireless technology, alongside with Vehicular Ad Hoc Networks (VANETs), brings exciting possibilities in terms of real-time data sharing, scalability, capability to integrate seamlessly AI-based solutions. In this paper, we propose a new design for 5G-enabled VANETs integrated with edge intelligence to support the AI-powered dynamic routing. The proposed approach is a catalyst to make an exploration in effective edge computing and big data processing by Ensemble DQN (DDQN) with Spatio-Temporal GNNs (ST-GNN) for dynamic routing decision optimization based on the real-time traffic flow analysis, road occupancy situation on-the-go and driver behaviors. Based on the ultra-low latency and high bandwidth features of 5G-based Vehicle-to-Everything (V2X) communication, the system guarantees real-time data propagation and swarm decision making among vehicles. Experimental results show that the embedding of edge AI can increase routing efficiency, decrease end-to-end delay and improve Quality of service (QoS) in vehicular networks. The research conceptualizes the latter as a key way to build the foundation for smart transportation systems with improved safety, mobility and energy.