Road-Related Event Detection and Dissemination Through 5G-Based Vehicle-to-Network-to-Everything Communications

Area of Science:

• Intelligent Transportation Systems
• Wireless Communications
• Machine Learning for Road Safety

Background:

• Connected and automated vehicles require accurate road-event detection and rapid alert dissemination to nearby vehicles and remote stakeholders.
• Vehicle-to-everything (V2X) communication alone is insufficient for reaching remote entities, necessitating cellular technologies like vehicle-to-network-to-everything (V2N2X).
• Integrating machine learning (ML) for event detection into V2N2X pipelines poses challenges for meeting strict latency requirements.

Purpose of the Study:

• To investigate and experimentally evaluate vehicle-to-network-to-everything (V2N2X) communications for disseminating hazardous road event alerts.
• To develop a proof-of-concept (PoC) platform integrating real-time event detection with message dissemination to both local and remote recipients.
• To analyze the impact of 5G deployment types and message broker load on V2N2X dissemination latency.

Main Methods:

• Development and experimental evaluation of a proof-of-concept (PoC) platform for V2N2X alert dissemination.
• Integration of machine learning (ML) algorithms for on-board hazardous road event detection.
• Leveraging 5G cellular connectivity and publish/subscribe messaging protocols for data dissemination.
• Testing under different 5G deployment scenarios (private vs. commercial) and varying message broker load conditions.

Main Results:

• The proposed V2N2X framework successfully combines real-time road-event detection with message dissemination to diverse recipients.
• Experimental results indicate that dissemination latency is significantly influenced by the type of 5G network deployment (private vs. commercial).
• Message broker load conditions were identified as a critical factor affecting the overall dissemination latency in the V2N2X system.

Conclusions:

• V2N2X communication, utilizing 5G and publish/subscribe protocols, is a viable approach for disseminating road event alerts beyond local vehicles.
• The choice of 5G network deployment and the management of message broker load are crucial for achieving low-latency alert dissemination.
• Further optimization is needed to guarantee prompt message delivery under stringent latency constraints when ML inference is part of the pipeline.