Fuzzy and Deep Belief Network Based Malicious Vehicle Identification and Trust Recommendation Framework in VANETs

This work aims to implement a clustering scheme to separate vehicles into a cluster that is based on various parameters, such as the total number of relay nodes, mistrust value, experience-based trust, and recommendation based trust. The Fuzzy based trust recommendation policy (FBTRP) creates a trust model that enables a recommendation and trust-based communication between the relay nodes. The FBTRP policy computes the trust value of each relay/intermediate node using two parameters namely network density and distance factor of the relay. Based on this trust value obtained from the FBTRP policy, the secure relay nodes can be selected to forward the data packets from the source to the destination. The proposed trust-based mechanism uses a Deep Belief Network (DBN) to predict the vehicle's malicious behaviour in the future using the threshold value obtained and divide the vehicles into three lists namely Green (Normal), Ash (Abnormal), and Black (Malicious). This trust results derived from the FBTRP-DBN model helps to select an optimal cluster head using the highest trustworthy node present in the VANET network by eliminating the malicious ones. The main aim of this work is to prevent malicious attacks from the trusted node and prevent the VANET network from DoS attacks. The efficiency of the proposed FBTRP-DBN trust model is evaluated by comparing it with different state-of-art techniques using various performance evaluation metrics such as False Positive Rate, Detection Rate, Processing Delay, Network Accuracy, PDR, Throughput Ratio, End to End delay, Communication Overhead, and False Alarm Detection Rate.