The field of connected vehicles is experiencing significant growth, with an increasing number of features being introduced as the technology advances. A rise in the complexity of inter-vehicle functions accompanies this expansion in market penetration. The primary goal of these advancements is to achieve fully autonomous traffic systems that ensure physical safety and address privacy concerns, reliability, optimality, comfort, and environmental sustainability. Additionally, these advancements are expected to deliver economic benefits to society at large.
As vehicles become more interconnected through various V2X (Vehicle-to-Everything) technologies, including 802.11p, IEEE 802.11bd, LTE-V2X, C-V2X, 5G NR, and heterogeneous V2X, the complexity of their architecture increases. These technologies support a range of applications such as safety, traffic management, public safety, and infotainment. This growing complexity necessitates the development of sophisticated simulation tools to ensure that communication systems and algorithms are reliable and secure.
The communication between vehicles (V2V) and between vehicles and infrastructure (V2I) plays a critical role in optimizing the control of autonomous vehicles, especially at intersections. As these communication-based algorithms become more prevalent, these systems’ reliability, sensitivity, safety, and security become paramount. To address these challenges, detailed simulations are required to evaluate various factors such as radio resource management, noise, interference, latency, packet size, and overall reliability.
In our latest research, simulation tools like SUMO (Simulation of Urban Mobility) and OMNeT++ are employed to model vehicle dynamics and communication networks in high detail. SUMO focuses on vehicle and traffic modeling, while OMNeT++ provides a framework for network simulation. The Veins and INET frameworks further integrate these simulations to create a comprehensive test environment for connected vehicles. However, increasing the level of detail in simulations can degrade performance, particularly when simulating high traffic densities and detailed V2X communications.
To mitigate these performance issues, we explore approaches such as microscopic simulations with high detail and mesoscopic solutions that balance detail with speed. Mesoscopic simulations aggregate communication details for vehicles further from the EGO vehicle while retaining crucial communication factors like interference and signal propagation. This approach aims to maintain the accuracy of communication models while improving simulation efficiency.
Moreover, the frequency of standardized Cooperative Awareness Messages (CAM), which provide essential data on vehicle position and dynamics, is a vital factor in simulations. We can enhance simulation performance by approximating CAM frequencies using neural network function approximators and aggregating microscopic network nodes into a mesoscopic simulation. This method strikes a balance between detailed modeling and efficient execution, supporting real-time mixed-reality testing and rapid prototyping.
Overall, addressing the performance challenges of V2X simulations remains a crucial area of research. This includes developing methods to improve simulation speed while ensuring the accuracy of communication models, particularly in scenarios involving high traffic densities. The ongoing efforts to refine these methods will facilitate more effective testing and validation of connected vehicle technologies.
Read the full paper here: https://www.sciencedirect.com/science/article/pii/S1569190X24001175