Enabling Fairness-Aware and Privacy-Preserving for Quality Evaluation in Vehicular Crowdsensing: A Decentralized Approach

Abstract

With the rapid development of vehicular crowdsensing, it becomes easier and more efficient for mobile devices to sense, compute, and measure various data. However, how to address the fair quality evaluation between the platform and participants while preserving the privacy of solutions is still a challenge. In the work, we present a fairness-aware and privacy-preserving scheme for worker quality evaluation by leveraging the blockchain, trusted execution environment (TEE), and machine learning technologies. Specifically, we build our framework atop the decentralized blockchain which can resist a single point of failure/compromise. The smart contracts paradigm in blockchain enforces correct and automatic program execution for task processing. In addition, machine learning and TEE are utilized to evaluate the quality of data collected by the sensors in a privacy-preserving and fair way, eliminating human subject judgement of the sensing solutions. Finally, a prototype of the proposed scheme is implemented to verify the feasibility and efficiency with a benchmark dataset.