A Light-weight Blockchain Architecture for V2V Knowledge Sharing at Vehicular Edges

Abstract

Autonomous vehicles (AV) utilize various machine learning (ML) models for performing tasks such as pedestrian detection, charging station prediction, routing path prediction, intrusion detection, to name a few. To improve the robustness of such applications, decentralized collaboration is imperative in a vehicular network. Moreover, decentralized collaboration poses threats, such as forged message injection, forged identity, and repudiatory activities, as most vehicles are stranger to each other. In recent years, blockchain technology has been widely used in the vehicular network to enhance trust, provenance, and eliminate unauthorized access to the vehicular services. However, most of the existing vehicular blockchain suffers from issues related to scalability, efficiency, and transaction verification. In this paper, we propose a lightweight vehicular blockchain architecture for distributed model sharing to enhance trust, verifiability, and non-repudiation in distributed vehicular collaboration. We propose a novel PoVS-BFT protocol and an effective two-step transaction verification mechanism for model sharing applications. Finally, simulation results are presented to conform to the efficacy of our proposed architecture. Simulation results show that the proposed PoVS-BFT protocol can minimize the size of consensus committee up to 62.5% which in turn reduces communication complexity during the consensus process.