RTPT: A framework for real-time privacy-preserving truth discovery on crowdsensed data streams

Abstract

The rapid proliferation of mobile sensing devices has promoted the emergence of a novel sensing paradigm mobile crowdsensing (MCS) for Internet of Things (IoT). For the data sensing in various mobile devices carried by different users, the sensed data may contain a variety of noise and sensitive information of mobile users. In order to extract the optimal truth from the noisy sensing data without privacy breach, privacy-preserving truth discovery (PPTD) has been proposed recently. However, there are two limitations to apply the existing PPTD approaches in real-time MCS applications. First, it is often ignored that the correctness of estimated truths may be susceptible to the dropout of mobile devices. Second, large-scale PPTD MCS application cannot be deployed, especially in real-time scenarios, due to the inefficiency of heavy cryptographic operations and iterative truth discovery algorithm. In this paper, we design and implement a real-time PPTD framework for crowdsensed data streams. Our design is based on the incremental conflict resolution on heterogeneous data, which is very efficient in processing data streams. In this framework, a low-overhead secure summation aggregation protocol is customized between an untrusted aggregator and a sensing client to estimate the true value of the sensed object. In addition, it should be noted that our system only needs one service provider. Theoretical analysis demonstrates that our design accomplishes both privacy-preserving and failure-robust. Through the complexity analysis and concrete implementation, the results show that the framework can achieve better scalability.