Self-healing AIS with Entropy Based SVM and Bayesian Aggregate Model for the Prediction and Isolation of Malicious Nodes Triggering DoS Attacks in VANET

Vehicle ad hoc networks, or VANETs, are highly mobile wireless networks created to help with traffic monitoring and vehicular safety. Security risks are the main problems in VANET. To handle the security threats and to increase the performance of VANETs, this paper proposes an enhanced trust based aggregate model. In the proposed system, a novel adaptive nodal attack detection approach - entropy-based SVM with linear regression addresses the trust factor with kernel density estimation generating the trustiness value thereby classifying the malicious nodes against the trusted nodes in VANETs. Defending the VANETs is through a novel reliance node estimation approach - Bayesian self-healing AIS with Pearson correlation coefficient aggregate model isolating the malicious node thereby the RSU cluster communication getting secure. Furthermore, even a reliable node may be exploited to deliver harmful messages and requires the authority of both the data and the source node to be carried out by the onboard units of the vehicles getting the reports of incident. DoS attacks (Denial of Service) disrupting the usual functioning of the network leads to inaccessible network to its intended users thereby endangering human lives. The proposed system is explicitly defending the VANET against DoS attacks as it predicts the attack without compromising the performance of the VANET handling nodes with various features and functions based on evaluating the maliciousness of attacking nodes accurately and isolating the intrusion. Furthermore, the performance evaluations prove the effectiveness of the proposed work with increased detection rate by 97%, reduced energy consumption by 39% and reduced latency by 25% compared to the existing studies.