FIDC: A framework for improving data credibility in mobile crowdsensing

Abstract

Mobile crowdsensing has become a popular paradigm to collaboratively collect sensing data from pervasive mobile devices. Since the devices used for mobile crowdsensing are owned and controlled by individuals with unpredictable reliability, varied capabilities, and unknown intentions, data collected with mobile crowdsensing may be untrustworthy. In particular, a mobile crowdsensing system is subject to collusion attacks where a group of malicious participants collaboratively send fake information to mislead the system. Defending against collusion attacks requires stronger defense mechanisms not available in existing works. In this paper, we propose a new framework for improving data credibility, named FIDC, in mobile crowdsensing to alleviate the threats posed by collusion attacks. FIDC seamlessly integrates two types of correlations: the spatial correlation of sensing data and the correlation between sensing data and provenance knowledge. While both correlations have been adopted separately in previous crowdsensing systems, the exploitation of an joint effort in FIDC poses a special technical challenge to fine-tune the performance. Evaluated extensively with a public mobile crowdsensing data for temperature monitoring, FIDC outperforms existing methods with respect to false detection accuracy and overall data credibility.