Secure, Privacy Preserving, and Verifiable Federating Learning Using Blockchain for Internet of Vehicles

Abstract

Internet of Vehicles (IoV) has been sought as a solution to realize an Intelligent Transportation System (ITS) for efficient traffic management. Data driven ITS requires learning from vehicular data and provide vehicles with timely information to support a wide range of safety and infotainment ITS applications. IoV is vulnerable to multitude of cyber-attacks and privacy concerns. Federated learning (FL) is on the verge of delivering the collaborative learning by exchanging learning model parameters instead of actual data, which is expected to provide privacy in IoV. However, despite featuring an inherently secure and privacy-preserving framework, FL is still vulnerable to poisoning and reverse engineering attacks. Blockchain technology (BC) has already demonstrated a zero-trust, fully secure, distributed, and auditable information recording and sharing paradigm. In this article, we present a practical prospect of blockchain empowered federated learning to realize fully secure, privacy preserving, and verifiable FL for the IoV that is capable of providing secure and trustworthy ITS services.