Privacy-Preserving Truth Discovery for Collaborative-Cloud Encryption in Mobile Crowdsensing

Abstract

In mobile crowdsensing (MCS) system, a variety of sensors are required to operate together to glean and upload sensory data to clouds for processing. In real practice, truth discovery has been widely explored to find reliable information from various mobile devices. Under the requirement of MCS security, privacy-preserving truth discovery (PPTD) causes wide concern, which refers to discovering truthful information from these unreliable uploaded data while protecting users' private information. Although many PPTD mechanisms have been proposed, they can either not guarantee low communication from users to the cloud, or fail to realize fully strong privacy protect including sensing data privacy, weight privacy, intermediate privacy, and estimated truth privacy. This study designs a collaborative cloud encryption architecture. In this framework, we propose a new system architecture that adapts a two-cloud peer model while leveraging garbled circuit (GC). Users only need to transfer data to two clouds once, which realizes low users workloads while supporting dynamic users. The two cloud terminals cooperate to complete the weight update through the GC but execute the truth discovery algorithm, respectively. In this way, the noninteractive comes true and the users' workloads are transferred to the cloud server side. The weight update finish and weight privacy are fulfilled through GC, meanwhile the other intermediate values maintain strong privacy. At the same time, because the collaborative cloud architecture of the two clouds ensures the confidentiality of the truth value in the cloud and the homomorphism at the inquiry side, it ensures the strong privacy of the whole process from users to inquiries. The performance of this proposed method is verified through extensive evaluations.