Context-Aware and Privacy-Preserving Trust Management Framework for Internet of Vehicles: Formal Models, Adaptive Consensus, and Real-World Validation

Abstract

The Internet of Vehicles (IoV) requires robust trust management to enable safe, autonomous transportation in dynamic, large-scale environments. Existing approaches face key limitations, including centralized vulnerabilities, static reputation models, privacy risks, and scalability bottlenecks. This article presents a theoretically grounded, privacy-preserving trust management framework that integrates context-aware trust updates, adaptive consensus selection, and efficient homomorphic encryption. The proposed multilayered trust evaluation model (MLTEM) ensures formal convergence under adversarial conditions. At the same time, the dynamic consensus adaptation algorithm (DCAA) dynamically adapts validator selection policies, endorsement requirements, and trust-score computation rules based on real-time network states, operating above Hyperledger Fabric’s fixed Raft ordering service. Privacy is preserved using optimized Paillier encryption with selective field encryption, batching, and GPU acceleration. Validated through over 5000 simulation trials and a 240-h real-world deployment with 50 vehicles and ten RSUs, the framework achieves 98.2% trust accuracy, 150-ms latency, 29% energy savings, and 97.6% attack detection across various threat types. These results confirm the framework’s practical readiness, scalability to 15 000 vehicles, and compliance with privacy regulations, demonstrating practical deployment potential.