Dynamic Mobile Network Slicing Through Vehicular Traffic Analysis

Abstract

Network slicing has emerged as a transformative enabler in 5G networks, offering tailored communication services for diverse traffic types on shared network infrastructure. In the context of autonomous driving and smart mobility, the ability to dynamically prioritize and manage sensor data—ranging from high-bandwidth video streams to low-latency text and binary position and coordination messages—plays a pivotal role in ensuring safe and efficient operation. This paper proposes a dynamic mobile network slicing framework designed to analyse vehicular traffic and adapt slicing policies to optimize resource allocation for autonomous driving applications. By leveraging distributed and disaggregated 5G network architectures, the proposed solution ensures seamless propagation of slicing policies across radio access networks (RAN) and core systems building end-to-end network slices. Experimental evaluations in scenarios such as Automated Guided Vehicle (AGV)-assisted operations in industrial environments demonstrate significant performance improvements, including a reduction in packet loss from 65% to 0% under congested network conditions. The results highlight the potential of dynamic slicing to enhance communication reliability and performance in autonomous driving ecosystems, supporting the seamless exchange of diverse sensor data types.