Novel machine learning based authentication technique in VANET system for secure data transmission

Abstract

Adaptive transport technologies based on vehicular ad hoc networks (VANET) has proven considerable potential in light of the developing expansion of driver assistance and automobile telecommunication systems. However, confidentiality and safety are the vital challenges in vehicular ad hoc networks which could be seriously impaired by malicious attackers. While protecting vehicle privacy from threats, it is imperative to stop internal vehicles from putting out bogus messages. Considering these issues, a novel machine learning based message authentication combined with blockchain and inter planetary file system (IPFS) is proposed to achieve message dissemination in a secured way. Blockchain is the emerging technology which attempts to solve these problems by producing tamper proof events of records in a distributed environment and inter planetary file system used in the framework is a protocol designed to store the event with content addressability. Along with this combined technology, the source metadata information collected from the inter planetary file system is stored via a smart contract and uploaded to the distributed ledger technology (DLT). For performing event authentication, K-means clustering and support vector machine (SVM) classifier is employed in this framework. K-means clustering performs clustering of vehicles and it is marked malicious or not malicious. After clustering, support vector machine classifier detects the malicious event messages. By this way, the malicious messages are identified and it is dropped. Only the secure messages are forwarded in the network. Finally, our approach is capable of creating a safe and decentralized vehicular ad hoc network architecture with accountability and confidentiality through theoretical study and simulations.