Tensor-based ranking-hiding privacy-preserving scheme for cloud-fog-edge cooperative cyber–physical-social systems

Abstract

Users living in Cyber-Physical-Social Systems (CPSS) generate massive amounts of data every day. The CPSS data may imply some reliable rules that can help CPSS better provide highly reliable services to humans. Nevertheless, the high-level reliable rules are very difficult to be mined and formalized. Therefore, we propose a Cloud-Fog-Edge Cooperative Reliable CPSS (CFECRC) framework for possibly adding reliable rules into CPSS. Ranked data is an important type of data in CPSS. How to design a secure, accurate and efficient ranking-hiding privacy-preserving scheme is a key challenge in CFECRC framework. However, existing privacy-preserving methods still have various shortcomings in the trade-off among privacy-preserving, analytic accuracy, and computational efficiency for ranking-hiding. To address the shortcomings, we propose a Tensor-based Ranking-Hiding Privacy-Preserving scheme (TRHPP) for CFECRC framework. First, we construct a set of 5th-order tensors to synthetically model item, user, location, time and weather as a whole to enhance analytic accuracy. Second, we obfuscate CPSS data and hide data ranking based on the obfuscated data to strengthen privacy-preserving and decrease computational overhead. The experimental results show that our scheme significantly outperforms existing classical schemes in privacy-preserving, analytic accuracy and computational efficiency simultaneously. This further verifies the feasibility of our framework.