Privacy methods for virtual sensors in a Sensing-as-a-Service industry paradigm

Abstract

The Internet of Things has recently received a lot of attention from researchers who envision it as a solution to fully and digitally automate urban environments and provide a collection of innovative services. One concept that has been envisioned is Sensing-as-a-Service, which uses the Cloud of Things infrastructure. For the CoT to operate and properly provide SaaS, many authors propose computational abstractions and organizational architectures to deploy systems in a scalable manner. However, privacy issues are often overlooked by publications in this field. Furthermore, the Industry 4.0 aims to apply a wide variety of sensors to enable data driven models and processes, which could benefit greatly from a more robust privacy-keeping methodology.The goal of this paper is to list and compare privacy alternatives for Sensing-as-a-Service deployments. Essentially, it complements a generic and simplified CoT architecture from the literature, which focuses solely on the performance of sensing tasks and inner workings of the network, with alternatives that satisfy the general privacy needs from the system’s users.A set of privacy methods are extracted from the literature and used to exemplify three approaches: privacy on a sensor level, privacy in data aggregation and processing, and privacy through cryptography on the application level. They are compared in regard to advantages, disadvantages and the architectural layers in which they should be implemented. Finally, some guidelines are discussed for validating CoT architectures in future research, while taking privacy into consideration and using the presented schemes.