LAUSR

Intelligent vehicular cyber-physical systems can perform automatic traffic measurement, which provides critical information for transportation engineering. However, one of the biggest challenges in traffic measurement is to protect the vehicles’ location and trajectory privacy, which may be revealed from the recorded traffic data. Prior studies in traffic measurement only offer heuristic privacy protection but lack a precisely defined privacy model. This article proposes an efficient traffic estimator with differential privacy protection. In our design, each road-side unit communicates with the passing vehicles and records their presence in a privacy-preserving data structure. By performing probabilistic analysis on the anonymized records, the proposed method can precisely estimate the number of common vehicles passed by multiple given locations during the given measurement period. Through theoretical analysis, we prove that the proposed method can protect the trajectory privacy of the vehicles with $\epsilon$-differential privacy even when the point privacy has been leaked. We also evaluated our traffic estimator based on a real-world transportation traffic dataset. The evaluation results demonstrate that the proposed estimator can achieve high estimation accuracy and high-level privacy protection through controllable tradeoffs.