Deep Learning Driven Blockchain For Secure Communications In Vanets

Abstract

Vehicular Ad hoc Networks (VANETs) facilitate vital vehicle communication for newlineimproved traffic management and secure messaging. Operating without fixed newlineinfrastructure, VANETs use wireless multi-hop channels, but face security challenges newlinedue to decentralized transmission and static networks. To address these concerns newlineand potential attacks, a secure communication approach employing deep learning and newlineblockchain has been introduced. newlineIn contribution 1, the vehicle data is collected and clustered by using the improved newlineK-Means clustering protocol and the Cluster Head (CH) has been selected based on newlinethe Firefly optimization algorithm (FA). From the selected nodes, the data are gathered newlineand encrypted using the Enhanced RSA (ERSA) algorithm for securing the data and newlinehave been transmitted from Cluster Member (CM) to Cluster Head. The VANETs newlinedata are collected in the Roadside Units (RSU) blockchain based on its authentication newlinein contribution 2. The VANETs data from the registered vehicles are then reviewed newlinefor trust and saved in the block using the Practical Byzantine Fault Tolerance (PBFT) newlineconsensus process. The malicious nodes with trust scores lower than the threshold newlinevalue in each cluster at the blockchain are removed. In contribution 3, the vehicle data s newlineconfidence score, message reliability and net reliability are assessed. The Enhanced newlineQPBFT is then employed to create the block and remove the rogue nodes from the newlinenetwork. The Deep Learning technique is used to detect the harmful nodes for the newlineintrusion type. The suggested model s performance is evaluated using clustering, newlineoptimization, data security, networking, trust evaluation and deep learning parameters. newlineThe proposed improved K-Means algorithm reached around 97% accuracy at 0.14 newlinesec clustering time for processing 200 nodes. The optimization performance of FA newlineis evaluated based on optimization time of 0.0729 sec. The performance of ERSA newlinehas been compared based on the key generation, encryption, decryption time and total newlinetime with RSA and EL Gamal algorithms. The n