Scalable HLA Co-Simulations of Connected and Automated Vehicles using Aggregation of Virtual Federates

Automated Vehicles (AVs) aim to transfer driving responsibility from humans to the vehicle itself with the objective of enhancing vehicle safety by increasing awareness and reducing response time. At present, automation in vehicles is limited to advanced driver-assistance systems such as adaptive cruise control, which require constant human supervision. A key step toward full automation is realizing communication between vehicles and road infrastructure (Vehicle-to-Infrastructure, V2I) and communication between different vehicles (Vehicle-to-Vehicle, V2V). Connected and Automated Vehicles (CAVs) that leverage V2V and V2I technologies (together called Vehicle-to-Everything, V2X) have information beyond the capabilities of their local sensors which can lead to better informed, and therefore safer, decisions. However, CAVs are not yet available to the general public and additional research is required to understand how these vehicles could perform during the various stages of their potential deployment. One approach to study the performance of CAVs at scale is co-simulation, where a road network could be modeled together with a V2X communications network to evaluate the impact of CAV deployments on the transportation system. The IEEE High-Level Architecture (HLA) is a co-simulation standard that is well-suited for evaluating such complex scenarios but has scalability issues with respect to the number of processes required to run a co-simulation and the amount of data those processes may need to exchange at runtime. This paper presents a novel approach to create scalable co-simulations of road transportation networks containing CAVs and vehicle communication networks. The CAVs are aggregated into a single process while retaining distinct virtual identities in the HLA co-simulation, reducing the number of processes required to control the vehicles and the number of messages required to communicate between simulations. A case study is provided to demonstrate the approach using a co-simulation between the Simulation of Urban Mobility (SUMO) traffic simulator and the OMNeT++ network simulator.