Collaborative-trust approach toward malicious node detection in vehicular ad hoc networks

Malicious node detection in vehicular ad hoc network (VANET) has always been a research hot spot. An efficient misbehavior detection scheme is needed to avoid and reduce the factor of selfishness and maliciousness especially in the case where the selfish beacons exploit the medium. To disseminate honest data over VANET infrastructure, it is very essential for nodes to collaborate with each other during the process of message forwarding and to ensure the successful delivery of honest data over the network. However, using the fake identities in message forwarding process easily results in the dissemination forged data over the network. Therefore, most of the existing techniques for detection of misbehavior in VANET use collaborative-trust–reputation-based approaches to tackle the issue forged and fake data transmission. The objective of this research is to calculate the trust weightages of each vehicle over the network and to reduce the intensity of the malicious vehicular nodes in VANET. The proposed collaboration-based maliciousness detection mechanism comprises data trust module and reputation calculating module, which guarantees honest data communication and reduces the false positive rate of malicious vehicular nodes. The data trust module uses trust evaluation and reputation calculation model to decide whether the vehicle is trustworthy using vehicular behavior vector in the context of packet transmission. The vehicular trust authority uses collaborative approach to integrate several trust evaluation assessments about a particular vehicular node and formulate a complete trust assessment. The performance evaluation shows that the proposed scheme delivers more priority messages with high true positive rate and low false positive rate. Moreover, the experimental results show that Gaussian kernel function best suits our proposed model in comparison with other rationalities. In addition, proposed models are more vigorous in context of true positive rate than the existing schemes, i.e., Dempster–Shafer theory of evidence, majority voted model and Bayesian inference.