Reputation-based Miner Node Selection in Blockchain-based Vehicular Networks

Abstract

The 21st century has brought many technological revolutions in transportation systems, data security, and network security, among many other technology elements. With the rise of smart cities, transportation systems are also being intelligent. An intelligent transportation system requires proper vehicle network infrastructure, data security, and intelligent decision-making capacity. Similarly, the networking for such systems can be provided by vehicular edge computing or vehicular ad-hoc network, which provides a required architecture for the communication between vehicles. The data transferred in such networks can be subjected to various attacks, such as false data injection attacks. To secure the system against such attacks, we need a secure architecture that guarantees the data security being transferred and stored in the system. Blockchain technology has the potential to provide such features. However, blockchain technology has vulnerabilities, such as majority attacks where the malicious actor/s control most miner nodes' mining power. To prevent the malicious actor/s from controlling the blockchain, we need a mechanism to identify and remove the malicious node from the consensus process. Such attacks can be countered if we can select the nodes allowed to take part in the consensus process, which can be done with the help of the miner node selection mechanism devised to select only specific miner nodes from all the available pool of nodes for the consensus process. In this study, we have proposed reputation-based miner node selection, where we have used an artificial neural network to calculate the reputation value of each miner node. Additionally, the vehicular network's purpose is to communicate with other entities in the network and interchange event information. This study proposed a novel accident event detection and validation method for our blockchain-based system. This study successfully implemented the proposed reputation-based miner node selection in Hyperledger Fabric and the accident event detection and validation in the blockchain network. We have obtained an accuracy of 77.56\% and a false positive rate of 2.76\% in our reputation model. Similarly, the implementation of reputation-based miner node selection in Hyperledger fabric yielded a throughput of approximately 125 ms per transaction. Additionally, the novel accident event detection and validation method required a validation time of 22 ms for a system with 5 nodes and increased proportionally with the number of nodes as expected. Thus, the study's result supported our proposed architecture and methodology to remove malicious nodes from the consensus process using reputation-based miner node selection and provided support for this system's real-time applicability and the novel accident event detection and validation system.