Abstract
Mobile crowdsensing (MCS) offers a scalable approach for urban air quality monitoring, but raises significant privacy concerns due to the collection of sensitive location and environmental data. This paper proposes a differential privacy (DP) framework tailored for MCS-based air quality monitoring, balancing data utility and privacy guarantees. We design a local DP mechanism that perturbs sensor readings at the participant's device before transmission, ensuring that individual contributions are indistinguishable. A truth discovery algorithm is employed to aggregate noisy data while mitigating the impact of outliers. Extensive simulations using realistic urban mobility traces and air quality models demonstrate that our approach achieves strong privacy protection (ε ≤ 1.0) with less than 10% degradation in monitoring accuracy compared to non-private baselines. The framework also supports dynamic privacy budgets and participant incentives. Our results indicate that differential privacy is a viable solution for privacy-preserving urban crowdsensing, enabling effective air quality monitoring without compromising participant confidentiality.