Distributed intersection control based on Cooperative Awareness Messages

Autonomous intersection control received particular interest recently. Theoretically, such traffic light-free controls can greatly improve the throughput of intersections with autonomous vehicles. The objective of this paper is to show how even a simple intersection control algorithm is susceptible to communication imperfections through detailed modeling of the communication layers. Safety and sensitivity analyses were carried out using a distributed control architecture, where no central control units are present. The control relies only on data contained in standard Cooperative Awareness Messages. The vehicle ad-hoc network communication is simulated with multiple layers considering many critical factors like signal propagation, noise, channel load, and even faulty units. Simulation results suggest that autonomous intersection control is robust against minor faults or communication errors. On the other hand, significant communication breakdowns lead to collisions.