Kinematic Information Aided User-Centric 5G Vehicular Networks in Support of Cooperative Perception for Automated Driving

Abstract

The cooperative perception of the driving environment via the sharing of locally sensed information among automated vehicles plays a fundamental role in ensuring the basic safety of automated driving in complicated public traffic. However, demanding requirements ranging from high data rate and large user density to ultra-high reliability and low latency, are imposed on the 5G network, which is considered the key enabler of cooperative automated driving. In this paper, we propose a novel ultra-dense 5G vehicular network architecture, which features the kinematic information aided user-centric access, to address these requirements. In particular, distributed local access and application centers (LAACs) are designed to perform application implementation and access control collectively, such that the kinematic information of the vehicles extracted at the application layer can be exploited in the dynamic management of network resources to sustain consistently high-performance wireless communications between vehicles and their serving LAACs. Focusing on the uplink transmission of the periodic cooperative sensing messages (CSMs), the possible design of key elements in the kinematic information aided user-centric access, including access point association, radio resource allocation, and mobility support, are discussed. Issues brought about by the practical network deployment and constraints are also considered. In addition, a practical benchmarking access strategy set, which addresses both the reliability and the latency requirements of CSMs, is proposed and evaluated by simulation under the freeway and intersection scenarios.