Secure and Efficient Blockchain-Based Knowledge Sharing for Intelligent Connected Vehicles

The emergence of Intelligent Connected Vehicles (ICVs) shows great potential for future intelligent traffic systems, enhancing both traffic safety and road efficiency. However, the ICVs relying on data driven perception and driving models face many challenges, such as the lack of comprehensive knowledge to deal with complicated driving context. In this paper, we investigate cooperative knowledge sharing for ICVs. We propose a secure and efficient blockchain based knowledge sharing framework, wherein a distributed learning based scheme is utilized to enhance the efficiency of knowledge sharing and a directed acyclic graph (DAG) system is designed to guarantee the security of shared learning models. To cater for the time-intense demand of highly dynamic vehicular networks, a lightweight DAG is designed to reduce the operation latency in terms of fast consensus and authentication. Moreover, to further enhance model accuracy as well as minimizing bandwidth consumption, an adaptive asynchronous distributed learning (ADL) based scheme is proposed for model uploading and downloading. Experiment results show that the DAG based framework is lightweight and secure, which reduces both chosen and confirmation delay as well as resisting malicious attacks. In addition, the proposed adaptive ADL scheme enhances driving safety related performance compared to several existing algorithms.

I. Introduction

With the development of communications and artificial intelligence, Intelligent Connected Vehicles (ICVs) is receiving increasing interest as a promising technology for tackling the challenges faced by intelligent transportation systems [1], [2]. Served by ubiquitous Vehicle to Everything (V2X) technologies, ICVs is capable of sharing sensing data and driving information with other vehicles and traffic infrastructures to improve traffic safety and efficiency. Equipped with intelligent on-board modules, ICVs support intelligent vehicular applications and various levels of driving automation. Several advanced driving use cases (including Advanced Driving Assistance System (ADAS), extended sensors and cooperative driving) have been specified in the fifth generation (5G) standards. However, the perception and driving models behind the ICVs are mainly data driven (e.g., trained by huge sensing and driving data). They are lack of knowledge to deal with unseen and complex driving scenarios. To address these problems, we are motivated to extend the sharing of raw sensing data to the sharing of knowledge for ICVs.