A Differentially Private Unscented Kalman Filter for Streaming Data in IoT

Abstract

With the evolution and development of the 5th generation (5G) technology, Internet of Things (IoT) within 5G provides a foundation and opportunity for smart home and smart healthcare. However, these scenarios will critically rely on the large-scale deployed sensors that constantly transmit streaming data to cloud platform for real-time estimation, which brings the issue of privacy disclosure since smart devices will gather kinds of data including personal sensitive information. Meanwhile, there is not some universal method to solve privacy problem in 5G because of diversified security needs for different applications. In this paper, an unscented Kalman filter based differentially private steaming data share scheme is proposed to protect user privacy for cloud platform in IoT. The proposed method can ensure that released data will not compromise individual privacy, and improve the utility of released data simultaneously. The proposed scheme is evaluated by four real-world datasets and compared with the results of utility optimization scheme based on Kalman filter. Experiments show that the proposed scheme enhances the utility of released streaming data under the premise of effective privacy preserving and achieve better practicability and validity.