A probabilistic model to corroborate three attacks in vehicular ad hoc networks

As VANET can be considered as a subclass of wireless ad hoc networks, they inherit the security problems. In this work, we focus on network security. We present some examples of attacks on these networks and we propose a method to optimize the security against three of these attacks. To improve the method of attacks' corroboration, we propose an innovative probabilistic model based on Logistic regression. This method will estimate the occurrence of an event (in this case, an attack). The method is based on a history of a knowledge base that estimates the attack's occurrences. When the regression model is validated, it will be used to estimate the probability of an attack and if it exceeds the threshold defined in advance, the attack is then confirmed.