Blockchain-Based Access Control Framework for Autonomous Vehicle Data Sharing

Abstract

Autonomous vehicles (AVs) generate vast amounts of data from onboard sensors, internal systems, and their surrounding environments. Sharing this data, particularly in stored form, is essential for enhancing safety, enabling post-event analysis, improving vehicle performance, and supporting regulatory oversight. However, the sensitive nature of AV data introduces significant challenges related to security, ownership, trust, and potential conflicts of interest among multiple stakeholders. Addressing these challenges requires a robust and transparent data-sharing framework that ensures traceability, accountability, and controlled access. This research investigates blockchain technology as a foundation for secure and trustworthy AV data sharing. Blockchain offers decentralisation, immutability, and fine-grained access control, making it a strong candidate for managing sensitive automotive data. Among available platforms, Hyperledger Fabric (HLF) is selected due to its permissioned architecture, modular design, and support for customisable endorsement policies, which align with the governance requirements of the automotive ecosystem. The research presents the design, implementation, and evaluation of a multi-party data-sharing framework for AVs using HLF and smart contracts. An additional contribution is the reconfiguration of endorsement policies (EPs), which define the participants required to validate transactions, to reflect real-world trust hierarchies. In particular, greater endorsement authority is assigned to vehicle manufacturers, reflecting their central role in the data lifecycle. To assess the implications of this design, three EP configurations are implemented and tested under varying workloads, measuring throughput, latency, and transaction success rate. The evaluation demonstrates that system throughput is constrained by architectural bottlenecks, including peer processing capacity, single-orderer contention, and state database overhead, rather than endorsement logic alone. User concurrency and chaincode complexity are identified as key factors limiting scalability, with throughput reaching a stable ceiling under increasing transaction loads. Among the evaluated configurations, Approach 2 shows greater resilience under high workloads, while Approach 1 achieves lower latency under lighter conditions. Furthermore, stricter endorsement policies improve security and accountability but introduce measurable performance overheads, whereas relaxed policies enhance efficiency at the cost of reduced resilience and potential centralisation. These findings indicate that practical AV data-sharing systems need to consider the alignment of endorsement policy design, chaincode complexity, and infrastructure provisioning with expected workload characteristics. They also highlight that scalability limitations in Hyperledger Fabric-based systems are driven primarily by system-level constraints rather than policy configuration alone, emphasising the need for broader architectural optimisation. Overall, this thesis presents a practical prototype for AV data sharing that integrates blockchain’s capabilities with configurable governance mechanisms, offering valuable insights for the design of data-sharing systems.