PrivacySphere: Privacy-Preserving Smart Spaces

Abstract

In smart spaces, data flows from sensors through data processing pipelines that interpret and enrich it to realize the needs of diverse applications. Smart space data may also be stored for future analysis and processing to implement new functionalities and learn correlations that can help improve deployed applications. Data processing may be performed at the edge (on sensors or at trusted local servers) or may be relegated to the (possibly untrusted) public cloud. This paper presents PrivacySphere, our vision towards a plug-n-play framework to integrate and test a variety of Privacy Enhancing Technologies (PETs) in smart spaces. PrivacySphere will support mechanisms (with appropriate APIs) to control when data is collected and from which sensors; in what format the data is exposed to devices/machines; and to whom (i.e., individuals/entities). Using PrivacySphere, flow of data may be intercepted at any point of the execution to apply PETs (e.g., differential privacy, encryption, policy-based sharing).