In VANET communication environments, denial-of-service attacks that compromise availability can cause more than just network problems; they can cause severe traffic jams or even traffic accidents that can lead to loss of life. Recognizing this risk, researchers are exploring various countermeasures. In particular, advances in machine learning and deep learning technologies have led to active research on dataset-based intrusion detection systems.
However, data collection in real-world VANET communication environments is a very complex task and can be expensive and time-consuming. In addition, data collected from vehicular networks may contain details of drivers and vehicles, raising privacy concerns. These issues make it very challenging to build datasets for training and evaluating machine learning and deep learning to develop intrusion detection systems.
Therefore, this paper details the process of dataset generation through simulation to address the limitations of data collection in VANET communication environments. To generate intrusion data, simulation tools such as OMNeT++, SUMO, Veins, and INET are used to set up various attack scenarios and modify standard protocols or create new applications to reflect the characteristics of each attack. The raw data generated through simulation is then preprocessed and transformed into a form suitable for training and evaluation of machine learning and deep learning-based intrusion detection models.
The validity of the generated dataset was verified by using various machine learning and deep learning models to accurately identify the attack patterns. The evaluation showed high detection rates of around 97% or more for most of the machine learning models. These results will play an important role in improving safety and security in VANET communication environments.