Enabling privacy by anonymization in the collection of similar data in multi-domain IoT

Abstract

The efficiency of Internet of Things (IoT) systems heavily relies on possessing trustworthy and up-to-date data. Thus, data collection represents a key feature in IoT applications and services that rely on data-streams. Reusing the same data-stream for distinct queries and services represents a promising way to save energy and reduce operational cost for the deployment. However, acquiring data from different owners through distinct providers incurs privacy issues. Indeed, data crosses the borders of different systems, and the end consumer may get knowledge on the original producer. In-network data aggregation enables the compression of information and improves the privacy while reducing the network load. However, the same data should not be aggregated several times, to provide correct and consistent operations. In this paper, we provide a pub/sub routing scheme for IoT systems that respects privacy constraints. The routers publish offers that describe the data they can export: they may aggregate several offers to respect the privacy constraints as defined by the producers. Each producer has a disposable ID, only used to detect and avoid overlaps in the data-streams. Our performance evaluation highlights the efficiency of our pub-sub routing solution to construct a valid aggregation (without overlap) distributively, while respecting privacy constraints.