Privacy-Preserving Truth Discovery for Sparse Data in Mobile Crowdsensing Systems

Feng Liu,Bin Zhu,Shaoxian Yuan,Jian Li,Kaiping Xue

DOI: https://doi.org/10.1109/globecom46510.2021.9685134

2021-01-01

Abstract:Truth discovery is an effective method to infer truthful information from a large amount of sensory data in mobile crowdsensing systems. Privacy-preserving truth discovery schemes require the cloud server not to access each worker's sensory data directly so that the privacy of sensory data can be preserved. In some specific applications such as sparse mobile crowdsensing, workers can only contribute sensory data on a small part of sensing tasks, implying that the information of which tasks are completed by a worker should also be preserved. However, existing privacy-preserving truth discovery schemes do not consider such sparse data scenarios in mobile crowdsensing systems. In this paper, we first identify the privacy issues in truth discovery when sensory data are sparse. To address these issues, we design a privacy-preserving truth discovery scheme by employing the additively homomorphic cryptosystem and additive secret sharing with two non-colluding servers. Through detailed analysis and extensive experiments, we demonstrate that our proposed scheme can satisfy strong privacy-preserving requirements with low computation and communication overhead.