Sybil Attack Detection in VANETs Using CatBoost Classifier

In the context of Vehicular Ad Hoc Networks (VANETs), Sybil attack, which makes fake identities in the network to spread false information between the network nodes, is becoming a major concern and real obstacle to the deployment of vehicular network. To address this problem, it is essential to have an efficient method for distinguish between legitimate nodes and fake nodes created by a malicious node. The CatBoost Model was trained on the Veremi Extension Dataset, using binary classification to predict the behaviors of a fake node. In this paper we evaluate the efficiency of a gradient boosting approach for Sybil attack detection using the open source dataset Veremi Extension. The preprocessing and cleaning the data is taken into consideration in the model training process, in order to improve the accuracy and performance of our model. Extensive simulation results confirm the efficiency of our model with a detection accuracy up to 98,60%.