On the Predictability of the Best V2X Path for Infrastructure-Assisted Automated Driving

When upcoming connected and automated vehicles (CAVs) will start their journey towards a given destination, one of the main criteria for the selection of their path may be the quality of service (QoS) that they expect to experience from the network along the traveled roads to support their driving functions. A long-term prediction of the QoS can be achieved using radio environmental maps (REMs) that however inherently average out the fast variations of the wireless channel. This paper focuses on comparing the long-term QoS prediction with the actual performance experienced by the vehicle along its path. For the former one, Sionna RT is used offline to generate REMs with specific granularity while averaging out fading effects. Sionna RT is then also used online, to produce channel realizations per each specific position of the vehicle. Simulation results comparing the performance achieved through the long-term channel prediction against the actual channel conditions, whose knowledge is provided online by Sionna RT, demonstrate that although precise point-by-point matching is not feasible, the overall QoS can be predicted with acceptable accuracy. Furthermore, an estimation is performed of the additional resources required to safely accommodate the strict requirements of a CAV application.