Advancing cooperative driving: from information freshness to large-scale and personalized platooning

Cooperative driving applications such as intersection collision avoidance and vehicular platooning promise safer, more efficient, and more sustainable road traffic. Yet, they still face open challenges that hinder practical large-scale deployment. On one hand, cooperative driving applications depend on timely information exchange, often over a shared wireless channel with finite capacity. As freshness (age of information, AoI) requirements vary across applications and scenarios, unrestricted beaconing is not viable, raising the question of how to ensure information freshness in a resource-efficient and context-aware manner. On the other hand, while platooning has been shown to be technically feasible, it remains unclear how to form platoons of passenger cars with unknown and heterogeneous trips and preferences, while also ensuring quantifiable personal benefits to incentivize participation. This thesis addresses these challenges with three sets of contributions: First, we propose a context-aware weighting model for the AoI metric that incorporates spatial relevance into the freshness evaluation. We apply this model in an adaptive beaconing mechanism to dynamically adjust transmission rates based on perceived importance, reducing wireless resource usage while maintaining desired information freshness for relevant data. Second, we formalize vehicle-to-platoon assignments as a similarity-based optimization problem and present three solution approaches: a globally optimal numerical solver and two greedy heuristics, including a fully distributed one that relies solely on local vehicle information. This method achieves near-optimal results with minimal assumptions and complexity, offering strong scalability and robustness. Third, we develop a cost-based platoon formation algorithm that employs a new metric to quantify total trip cost by combining fuel and time expenditures into a single economic value. It forms platoons only when doing so reduces individual driver costs, outperforming both standard adaptive cruise control and similarity-based methods across diverse traffic conditions and driver preferences. To support this work, we developed PlaFoSim, an open-source simulation framework for large-scale platooning studies. Collectively, these contributions advance cooperative driving systems toward practical deployment by emphasizing context awareness, resource efficiency, decentralized decision-making, and user-centered benefits.