Edge-based differential privacy computing for sensor–cloud systems

Abstract

In sensor–cloud systems, with more personal data being hosted in cloud, privacy leakage is becoming one of the most serious concerns. Privacy computing is emerging as a paradigm to systematically enhance privacy protection. In other words, the new paradigm requests us to improve the computing model to provide a general privacy protection service. In this paper, we propose an edge-based model for data collection, in which the raw data from wireless sensor networks (WSNs) is differentially processed by algorithms on edge servers for privacy computing. A small quantity of the core data is stored on edge and local servers while the rest is transmitted to cloud for storage. In this way, the benefits are twofold. First, the data privacy is preserved since the original data cannot be retrieved even if the data stored in the cloud is leaked. Second, implemented by a differential storage method, compared to the state of the art, the edge-based model sends less data to the cloud and reduces the cost of communication and storage. Both theoretical analyses and extensive experiments validate our proposed method.