Vehicular networks are distributed, self-organizing and highly mobile ad-hoc networks. Communication among networked cars is presumably one of the most important possibilities to enhance traffic safety. To achieve this, vehicles should be able to manage information like a black ice warning in a distributed manner. In addition, the publish/subscribe (Pub/Sub) paradigm allows for information filtering and tracking of changes in the information environment. For example, warnings could be sent to affected vehicles only or vehicles could track traffic changes. The approach presented in this paper utilizes structured peer-to-peer (P2P) mechanisms for distributed information management. To adapt to the highly mobile environment, the vehicular network is divided into segments which form separate, interacting P2P networks. The Pub/Sub functionality is built on top of these segments. This way, P2P-based Pub/Sub can be applied with no need to maintain an event distribution tree. The proposed approach enables vehicle-to-vehicle applications to publish information referring to certain areas or validity durations. Moreover, published information can be modified and deleted and solely interested vehicles can be notified about these events.

Today, a huge variety of sensors is built into vehicles enabling them to obtain a view of their surroundings based on locally observed data. With VANETs (vehicular ad-hoc networks) as an emerging technology, it is possible both to disseminate data to other vehicles and to collect information in order to improve their own picture of the current situation. However, distributed applications based on VANETs still need to agree on a "common understanding" of context for interoperability on a contextual level. In recent years, ontologies have become the "silver bullet" for describing concepts and their relationship in a semantically rich manner. In this paper, we apply this approach to context modeling and propose a collision avoidance application in order to show that applications can benefit from the advantages of ontological models like distributed composition, partial validation, richness and quality of information and a certain level of formality. Also, we want to prove that calculations on the model are still fast enough to fulfill real-time requirements imposed by active safety for vehicles. Two different scenarios with different context models but unchanged applications show the advantages of ontologies with respect to extensibility of models. In addition, it is shown that larger and more complex models do not necessarily increase the execution times.