Study of Masquerade Attack in VANETs with Machine Learning

Vehicular Ad Hoc Network (VANET) affords communication between vehicles and roadside infrastructures in order to improve road safety and driving conditions. The implementation of precise security mechanisms is imperative in VANET to ensure safety, as attackers are constantly seeking ways to exploit network vulnerabilities. Among the prevalent risks, the masquerade attack stands out as a common and impactful threat. Masquerade is performed by malicious nodes to gain access to the network by impersonating the real identity of a legitimate vehicle. In this paper, we evaluate three variants of masquerade with different machine learning algorithms. These experiments are realised using the F2MD simulation environment with Weka. The performance of the misbehavior detection mechanisms is evaluated by accuracy, recall, and precision. Our comparative results demonstrate that the Random Forest (RF) algorithm outperforms the others in the proposed simulation scenario in terms of accuracy, while the decision tree algorithm (J48) excels as the fastest with minimal prediction time.