A Privacy-Preserving Reputation-Based Incentive Scheme for Crowdsourced Map Updates in Autonomous Vehicles

Abstract

The dynamic maintenance of maps via Mobile Crowdsensing (MCS) is critical for autonomous driving but faces dual challenges regarding data trustworthiness and user privacy. Specifically, malicious vehicles may launch strategic attacks to degrade map accuracy, while traditional incentives often leak user trajectories through reward flows. To address these issues, we propose Privacy-preserving Reputation-based Incentive Scheme for Maps (PRISM), a cloud-assisted privacypreserving framework for trustworthy map updates. The main contributions are threefold: 1) We design an authentication protocol using group signatures and Task-Specific Pseudonyms (TSP), offering user anonymity toward the platform server and conditional traceability via the trusted authority. 2) We propose a reputation-initialized robust truth discovery algorithm integrated with a physics-aware scoring function, which supports unlinkable task-level participation records under the anonymous reputation token. 3) We devise an adaptive reputation penalty mechanism to defend against strategic On-Off attacks and utilize Key-Derived Partially Blind Signatures (KD-PBS) to achieve tamper-proof and unlinkable reward issuance. Extensive simulations based on T-Drive-derived semi-synthetic traces demonstrate that PRISM significantly outperforms state-of-the-art baselines in terms of localization accuracy, robustness, and operational efficiency.