A Reliable and Privacy-Preserving Truth Discovery Scheme for Mobile Crowdsensing Based on Functional Encryption

The rapid proliferation of human-carried devices has facilitated the emergence of data collection paradigm called mobile crowdsensing. However, data collected from multiple sources may not be reliable and may contain privacy-sensitive information. To resolve these issues, privacy preserving truth discovery has been proposed as an effective mechanism to protect the privacy sensitive information while the truth is computed. Nevertheless, most existing works do not meet the requirements of full privacy protection and are not robust against malicious sensory data injection, as in these schemes outliers or data encrypted with wrong public key will greatly impact accuracy of the truth. In our paper, we present a reliable and privacy preserving truth discovery scheme (FE-TDS) which utilizes a combination of multi-client functional encryption (MCFE) and data perturbation technique to address the privacy issues. Moreover, we design two ciphertext-based filtering methods to remove outliers and data encrypted with wrong public key in a private manner. Detailed theoretical analysis has been conducted to demonstrate the security and feasibility of our scheme. Furthermore, external experiments have been carried out to demonstrate the effectiveness and practicality of our scheme.